CN117522953A - An automatic coffee particle size assessment method and related equipment - Google Patents

Abstract

The invention discloses a coffee particle size automatic evaluation method and related equipment, wherein the method comprises the following steps: identifying a central circular area of an original coffee particle size image by adopting a Hough circle detection algorithm to obtain the coffee particle size image, denoising and gray processing the coffee particle size image to obtain a gray image, and converting the gray image into a binary image; carrying out contour boundary identification on coffee particles in the binary image, and extracting the contour of all communicated edges to obtain a contour boundary; smoothing the profile boundary to remove saw teeth; obtaining the outer coating of all particles according to the convex hulls of the particles; removing silver skin in the binary image; dividing the adhered coffee particles in the binary image to obtain a recognition result of the coffee particles; and obtaining the particle size evaluation result of the coffee particles according to the identification result and the corresponding particle size and specific surface area statistical information. The invention can greatly reduce the detection cost and realize accurate and efficient evaluation of the particle size and specific surface area of the coffee particles.

Description

An automatic coffee particle size assessment method and related equipment

Technical field

The invention relates to the technical field of coffee particle size detection, and in particular to an automatic coffee particle size assessment method, system, terminal and computer-readable storage medium.

Background technique

The particle size and specific surface area of coffee particles play a key role in the extraction process and are also important criteria for the study of coffee particles. After coffee beans are ground, extraction operations are required to extract soluble compounds from the coffee powder, such as caffeine, oils, organic acids and other components that are critical to the flavor and characteristics of coffee. In general, specific surface area is defined by the ratio of the area to volume of a particle, and the smaller the particle size, the greater the specific surface area, allowing more dissolved compounds to be released. Therefore, the particle size and specific surface area of coffee particles are crucial to adjusting the taste of coffee, and it is necessary to accurately and efficiently evaluate the particle size and specific surface area of coffee particles.

At present, the detection of coffee particle size mainly uses laser-induced piezoelectric technology. The use of laser-induced piezoelectric technology to detect coffee particle size is a non-contact measurement method that combines laser scattering and piezoelectric effects. To determine the size of the particles, this process can be roughly divided into the following steps: (1) Laser irradiation: coffee powder or particles are placed in an appropriate container, and the laser beam is accurately irradiated onto the coffee particles; (2) Detection of scattered light: When the laser beam hits the coffee particles, the light will be scattered. The angle and intensity of the scattered light depend on the size and shape of the particles. One or more light detectors are used to capture the scattered light; (3) . Signal conversion: The detected scattered light signal is transmitted to a piezoelectric device. The piezoelectric device can convert the light signal into an electrical signal. The intensity and shape of the electrical signal are related to the characteristics of the scattered light, and thus related to the size of the coffee particles. ; (4) Data analysis: transmit the electrical signal to a computer system for analysis, and use appropriate algorithms and software to calculate the size of coffee particles based on the characteristics of the electrical signal. The results can be displayed in the form of numbers or graphics; (5 ), Determination of particle size: Based on the analysis results, the size distribution of coffee particles can be determined.

In general, the disadvantages of using laser-induced piezoelectric technology are the high detection cost and the inability to identify silver skin, which makes it impossible to accurately and efficiently evaluate the particle size and specific surface area of coffee particles.

Therefore, the existing technology still needs to be improved and developed.

Contents of the invention

The main purpose of the present invention is to provide an automatic coffee particle size assessment method, system, terminal and computer-readable storage medium, aiming to solve the high detection cost of using laser-induced piezoelectric technology to detect coffee particle size in the prior art. , and the silver skin cannot be identified, resulting in the problem of being unable to accurately and efficiently evaluate the particle size and specific surface area of coffee particles.

In order to achieve the above object, the present invention provides a method for automatic evaluation of coffee particle size. The automatic evaluation method of coffee particle size includes the following steps:

Obtain the original coffee particle size image, use the Hough circle detection algorithm to identify the central circular area of the original coffee particle size image, and obtain the coffee particle size image. The coffee particle size image is denoised and grayscale processed to obtain the grayscale image. grayscale image, and convert the grayscale image into a binary image;

The edge tracking algorithm based on connectivity performs profile boundary recognition on the coffee particles in the binary image, extracts the contours of all connected edges, and obtains the profile boundary;

Smooth the outline boundary and remove jagged edges from the outline boundary;

The particle convex hull is obtained according to the smoothed binary image, and for each particle that has not obtained an outer envelope, processing is performed according to the particle convex hull until all particles obtain an outer envelope;

Confirm the particles belonging to silver skin according to the silver skin judgment conditions, and remove the silver skin in the binary image;

The particle segmentation algorithm based on the segmentation index segments the adhered coffee particles in the binary image to obtain the identification result of the coffee particles;

Based on the identification results of coffee particles and the corresponding particle size and specific surface area statistical information, the particle size evaluation results of coffee particles are obtained.

In addition, to achieve the above object, the present invention also provides an automatic coffee particle size evaluation system, wherein the automatic coffee particle size evaluation system includes:

The coffee image preprocessing module is used to obtain the original coffee particle size image, use the Hough circle detection algorithm to identify the central circular area of the original coffee particle size image, obtain the coffee particle size image, and remove the coffee particle size image. Noise processing and grayscale processing obtain a grayscale image, and convert the grayscale image into a binary image;

The coffee particle profile boundary recognition module is used to perform profile boundary recognition of coffee particles in the binary image using an edge tracking algorithm based on connectivity, extract the contours of all connected edges, and obtain the profile boundary;

A filtering and anti-aliasing module, used to smooth the outline boundary and remove the aliasing of the outline boundary;

The particle outer envelope acquisition module is used to obtain the particle convex hull according to the smoothed binary image. For each particle that has not obtained the outer envelope, it is processed according to the particle convex hull until all particles obtain the outer envelope. ;

The coffee silver skin removal module is used to confirm the particles belonging to silver skin according to the silver skin judgment conditions and remove the silver skin in the binary image;

The particle adhesion segmentation module is used to segment the adhered coffee particles in the binary image using a particle segmentation algorithm based on the segmentation index to obtain the identification result of the coffee particles;

The particle size assessment module is used to obtain the particle size assessment results of coffee particles based on the identification results of coffee particles and the corresponding particle size and specific surface area statistical information.

In addition, to achieve the above object, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and an automatic coffee particle size assessment program stored on the memory and executable on the processor, When the coffee particle size automatic assessment program is executed by the processor, the steps of the coffee particle size automatic assessment method as described above are implemented.

In addition, to achieve the above object, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an automatic coffee particle size evaluation program. When the coffee particle size automatic evaluation program is executed by a processor, Steps to implement the automatic coffee particle size assessment method as described above.

In the present invention, an original coffee particle size image is obtained, a Hough circle detection algorithm is used to identify the central circular area of the original coffee particle size image, and a coffee particle size image is obtained. The coffee particle size image is denoised and grayed out. The grayscale image is obtained through degree processing, and the grayscale image is converted into a binary image; the edge tracking algorithm based on connectivity performs contour boundary recognition on the coffee particles in the binary image, and extracts the contours of all connected edges. , obtain the profile boundary; smooth the profile boundary to remove the jaggedness of the profile boundary; obtain the particle convex hull according to the smoothed binary image, and obtain the outer envelope for each particle. , process according to the convex hull of the particles until all particles obtain the outer envelope; confirm the particles belonging to the silver skin according to the silver skin judgment conditions, and remove the silver skin in the binary image; the particle segmentation algorithm based on the segmentation index The adhered coffee particles in the binary image are segmented to obtain the identification results of the coffee particles; based on the identification results of the coffee particles and the corresponding particle size and specific surface area statistical information, the particle size evaluation results of the coffee particles are obtained. The present invention detects coffee particles based on image vision, which can greatly reduce detection costs and achieve accurate and efficient evaluation of the particle size and specific surface area of coffee particles.

Description of drawings

Figure 1 is a flow chart of a preferred embodiment of the automatic coffee particle size assessment method of the present invention;

Figure 2 is a schematic diagram of the overall principle of realizing coffee particle size assessment in a preferred embodiment of the automatic coffee particle size assessment method of the present invention;

Figure 3 is a schematic diagram of the coffee particles ground by the grinder being placed on a circular lamp and photographed by a professional camera in a preferred embodiment of the automatic coffee particle size assessment method of the present invention;

Figure 4 is a schematic diagram of the coffee particle identification effect obtained by obtaining the outer envelope of the particles in a preferred embodiment of the automatic coffee particle size assessment method of the present invention;

Figure 5 is a schematic diagram of the segmentation effect of coffee adherent particles in a preferred embodiment of the automatic coffee particle size assessment method of the present invention;

Figure 6 is a schematic diagram of particle size distribution in a preferred embodiment of the automatic coffee particle size assessment method of the present invention;

Figure 7 is a schematic diagram of the specific surface area distribution in a preferred embodiment of the automatic coffee particle size assessment method of the present invention;

Figure 8 is a schematic diagram of the principle of a preferred embodiment of the automatic coffee particle size assessment system of the present invention;

Figure 9 is a schematic diagram of the operating environment of the preferred embodiment of the terminal of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention clearer and clearer, the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described here are only used to explain the present invention and are not intended to limit the present invention.

The automatic evaluation method of coffee particle size according to the preferred embodiment of the present invention is as shown in Figure 1 and Figure 2. The automatic evaluation method of coffee particle size includes the following steps:

Step S10: Obtain the original coffee particle size image, use the Hough circle detection algorithm to identify the central circular area of the original coffee particle size image, obtain the coffee particle size image, and perform denoising and grayscale processing on the coffee particle size image. The grayscale image is processed and converted into a binary image.

Specifically, as shown in Figure 3, after the coffee is ground by a grinder, the resulting particles are placed on a circular lamp and photographed using a professional camera (such as a CMOS camera) to obtain the original coffee particle size image collected by the CMOS camera. (that is, the original image), using the Hough Circle Transform algorithm (Hough Circle Transform), which is an algorithm used to detect circular objects in images. It does not directly rely on the brightness of the central area, but is based on the geometric characteristics of the circle. Detection is performed, that is, used to detect circular areas) to identify the central circular area in the original coffee particle size image whose brightness is higher than the preset brightness (that is, to identify the central area with higher brightness in the center of the circular lamp), Get an image of coffee particle size.

In the process of obtaining the image, due to the influence of various factors, the obtained image contains some noise information, resulting in deterioration of image quality, so denoising processing is required; the present invention uses a Gaussian filter to perform denoising on the coffee particle size image. Denoising process, if the two-dimensional Gaussian function is:

Among them, σ is the standard deviation of the Gaussian function, x and y represent the abscissa and ordinate of a certain pixel in the original coffee particle size image. By comparing the rows and columns of the original coffee particle size image f (x, y) with Gaussian The function G(x,y) is convolved to obtain a smooth image K(x,y); noise can be suppressed through smoothing filtering, and the standard deviation of the Gaussian function can control the degree of smoothing. When the standard deviation is small, the Gaussian filter is relatively short, the convolution calculation is small, and the accuracy is high, but the signal-to-noise ratio will also be reduced and the smoothness is poor; when the standard deviation is large, the convolution calculation amount increases. , the accuracy is low, but the loudness signal-to-noise ratio is high and the smoothness is good.

Considering that brightness can best reflect the difference between coffee particles and white background, and the RGB color space is difficult to describe the "distance" between colors, in order to facilitate the subsequent determination of whether each pixel in the picture belongs to the background color or coffee particles, it is considered to convert the color image into Perform grayscale processing.

The denoised smoothed image K(x,y) is grayscale processed using the weighted average method to obtain a grayscale image. The formula is:

I=αR+βG+γB;

Among them, R, G, and B are the three components of a certain pixel of the original coffee particle size image, α, β, and γ are intensity coefficients, and satisfy α+β+γ=1, and I represents the grayscale image.

Finally, after filtering, the grayscale image is binarized and converted into a binary image. The binary image is used to characterize the coffee particles and background color. For example, draw a histogram of the grayscale image and observe the grayscale image. For the pixel value of the edge point of the coffee particles, select an appropriate grayscale threshold δ. The grayscale threshold δ is a set grayscale used to convert the grayscale image into a binary image. All grayscale values in the image are greater than or Pixels equal to δ will be set to white (or 1), and all pixels with a grayscale value less than δ will be set to black (or 0), resulting in a binarized image.

Step S20: The edge tracking algorithm based on connectivity performs profile boundary recognition on the coffee particles in the binary image, extracts the outlines of all connected edges, and obtains the profile boundary.

Specifically, the present invention uses a connectivity-based edge tracking algorithm to identify the contour boundaries of coffee particles. This algorithm can extract the contours (profile boundaries) of all connected edges from a binary image.

The basic idea of the algorithm is to start from a certain starting point and gradually track along the edge direction until returning to the starting point. During the tracking process, the position of the next point to be tracked is determined based on the state of the current point's domain pixels. The specific steps of the algorithm are as follows:

Step 1: Select any white pixel in the binary image as the starting point;

Step 2: Starting from the starting point, scan the surrounding 8 pixels in a clockwise or counterclockwise direction to find a black pixel as the next point. If a black pixel is found, mark the black pixel as the current point. , and continue to the next scan. If no black pixel is found, the tracking ends;

Step 3: After finding the next point, mark the connection between the current point and the next point as an edge, and use the next point as the current point to continue the next scan;

Step 4: End tracking when returning to the starting point, and return all marked edge points as a contour;

Step 5: If there are multiple connected contours, start from an unvisited starting point and repeat the above steps 1-4 until all contours are extracted.

It should be noted that in order to avoid repeated scanning and endless loops, the algorithm will mark the visited pixels as visited when performing edge tracking to ensure that each pixel is only visited once. Obtain the contour boundary. After obtaining the contour boundary, record the length _i of the pixel boundary of the i-th particle and the mean mean _i of the gray value of all pixels of the i-th particle.

Step S30: Smooth the outline boundary and remove jagged edges from the outline boundary.

Specifically, since the directly obtained profile data consists of a large number of dense point sequences, the obtained profile boundary is jagged. On the basis of implementing the above profile boundary recognition algorithm, the profile boundary needs to be smoothed.

The present invention uses moving average filtering (moving average filtering is a commonly used signal processing technology, its main purpose is to reduce random fluctuations in the data, thereby more clearly revealing the trend or characteristics of the data) to smooth the profile boundaries of coffee particles. , in the process of average filtering, obtain a suitable (error minimization), effectively eliminate or reduce random fluctuations and noise in the data, strike a balance between denoising and retaining signal trend information, and be able to achieve a reasonable time React to signal changes within, appropriate filter length) moving average filtered output signal Make the output signal The error between each measured value x(u) (i.e. input signal) within a sampling period/> The sum of squares is the smallest, where i=1, 2,...,N, then there is:

N ^* =argmin(W);

Find the extreme value and get:

Among them, N represents the number of points used in moving average filtering, W represents an objective function that needs to be minimized, and N ^* represents the point or set that minimizes the objective function W.

Since the acquired profile data is annular data connected end to end, in order to achieve symmetric mean (symmetric mean moving average filter is a special type of moving average filter, it considers the data points before and after the current point when processing the data, To maintain the symmetry of the data. This means that the filter will use the current data point and its surrounding data points to calculate the average to determine the filtered output) The moving average filter can process both endpoints, and first needs to The profile array is spliced, and then moving average filtering is performed. If there are n data points at the starting profile boundary, the point set is {(x ₁ , y ₁ ), (x ₂ , y ₂ ), ..., (x _{n -1} ,y _n-1 ), (x ₁ ,y ₁ )}, recorded as:

Take the last (n-1) row sub-matrix of matrix A as:

Splicing matrix A and sub-matrix B, we get:

Perform moving average filtering on the column vectors of matrix C respectively:

when get After the calculation result of one of the points, the subsequent filtering result is calculated using the pre-order result:

Among them, p=(N-1)/2, q=p+1;

Record the filtered matrix as:

Get matrix The n-row submatrix D is the filtering result of the silhouette boundary point set:

Among them, i=n/2+1, represents the average value of the measured value x(i); finally, each contour boundary that has not been filtered and de-aliased is filtered in turn, and the above steps are repeated in sequence until all contour boundaries are smoothed.

After identifying the edges of the coffee particles, the average filtering algorithm is used to smooth the edges of the particles, which can improve the calculation accuracy of the coffee particle size and is helpful for removing silver skin and adhesion segmentation of the particles.

Step S40: Obtain the particle convex hull according to the smoothed binary image, and perform processing according to the particle convex hull for each particle that has not obtained an outer envelope until all particles obtain an outer envelope.

Specifically, in the captured image, it is inevitable that multiple particles are in contact with each other and are considered to be single particles after binarization, thus affecting the statistical results. The prerequisite for separating these adherent particles is to obtain the convex hull of the particles. The convex hull is a convex polygon containing the smallest boundary points.

If P ₁ (x ₁ ,y ₁ ), P ₂ (x ₂ ,y ₂ ),..., P _n (x _n ,y _n ) are a set of n points on the smoothed contour boundary, where x Arrange from small to large. If x is equal, then y is arranged from small to large. Select the point P ₁ with the smallest abscissa as the starting point. If the abscissas are the same, select the point with the smallest ordinate as the starting point. According to geometry As a general rule, P ₁ must be the vertex of the convex hull of the set. It is known that P ₁ is a point on the convex hull, then:

Among them, i∈(2,n], if (n-2) results all have the same sign, then P ₂ is a point on the convex hull, otherwise, P ₂ is not a point on the convex hull.

If P ₂ is a point on the convex hull, then:

Among them, j∈(3,n], if (n-3) results of all have the same sign, then P ₃ is a point on the convex hull. Otherwise, P ₃ is not a point on the convex hull. P ₄ , P ₅ ,..., P _n are tested in the same way, Until the end of the traversal, all points that meet the convex hull conditions are found.

Finally, each particle that has not obtained an outer envelope is processed sequentially until all particles obtain an outer envelope.

For any polygon, the coordinates of each vertex are A ₁ (x ₁ ,y ₁ ), A ₂ (x ₂ ,y ₂ ),..., A _n (x _n ,y _n ), then the perimeter and area of the polygon are for:

Among them, x _n+1 =x ₁ , y _n+1 =y ₁ ;

Calculate the perimeter and area of the particle profile, respectively, and record them as P ₁ and S ₁ respectively; calculate the perimeter and area of the particle convex hull, and record them as P ₂ and S ₂ respectively;

Define the ratio of the actual area of the particle to the convex hull area as the shape coefficient, denoted as

After the above steps, the obtained coffee particle recognition effect is shown in Figure 4. (a) in Figure 4 represents the coffee particle edge detection result, (b) in Figure 4 represents the coffee particle edge smoothing effect, and Figure 4 (c) shows the outer envelope recognition result of coffee particles.

Step S50: Confirm the particles belonging to silver skin according to the silver skin judgment conditions, and remove the silver skin in the binary image.

Specifically, the particles in the acquired image are not all coffee particles, but some are silver skins on the surface of coffee beans. In order to eliminate the interference of silver skins on the experiment, the silver skins are judged based on the difference between the silver skins and coffee particle sizes, and the particle pixel length is set. Threshold α ₁ , α ₂ and mean threshold value β ₁ , β ₂ of particle pixel length.

Pre-defined silver skin judgment conditions include:

Silver skin judgment condition 1: The i-th particle boundary pixel length length _i > α ₁ and the mean mean _i > β ₁ ;

Silver skin judgment condition 2: The i-th particle boundary pixel length length _i > α ₂ and the mean mean _i > β ₂ ;

If the i-th particle satisfies any one of the silver skin judgment condition 1 and the silver skin judgment condition 2, the i-th particle is considered to be silver skin, and the silver skin is removed.

Step S60: The particle segmentation algorithm based on the segmentation index segments the adhered coffee particles in the binary image to obtain the identification result of the coffee particles.

Specifically, as shown in Figure 4, during the particle size analysis process, there may be two coffee particles adhering to each other. In order to solve the problem of coffee particles adhering, the present invention designs a particle segmentation algorithm based on segmentation index. The algorithm ideas and steps are as follows:

Step 1: The particle profile boundary has been obtained. If Q ₁ =(x ₁ ,y ₁ ), Q ₂ =(x ₂ ,y ₂ ),..., Q _n =(x _n ,y _n ) are the profile boundaries n points on the top, and x are arranged from small to large. If x is equal, then y is arranged from small to large, where Q _i = (x _i , y _i ), Q _j = (x _j , y _j ) are the contours Two different points (j>i) on the boundary of the shape, if the Euclidean distance D ₁ is the straight-line distance between the two points, then:

If the perimeter distance D ₂ is the shortest distance from point Q _i to point Q _j along the contour boundary, the perimeter distance D ₂ is approximately the difference in serial numbers, then:

Obtain the Euclidean distance D ₁ and perimeter distance D ₂ of any two points on the contour boundary; let the ratio ratio be:

ratio=D ₁ /D ₂ ;

There are two points on the particle profile boundary that minimize the ratio value. These two points are regarded as key points, and the minimum ratio value is used as the segmentation index, and is recorded as min{ratio}.

Step 2: Set the particle shape coefficient threshold γ, the segmentation index threshold ε, and define the particle segmentation principles including:

Principle 1 of particle segmentation: shape _factor of particles >γ and segmentation index min{ratio}<ε;

Particle segmentation principle 2: The key point number (j-i) that satisfies the particle segmentation principle 1>3;

If the particle satisfies both particle segmentation principle 1 and particle segmentation principle 2, the particle will be divided into particle 1 and particle 2, otherwise it will not be divided.

Step 3: Smooth the division of the two particles. If the pixel length of the boundary of particle 1 gra ₁ >4, then smooth the division of particle 1; if the pixel length of the boundary of particle 2 gra ₁ >4, smooth the division of particle 2. Smoothing is performed at the division points; the smoothing method uses the moving average filtering method introduced in step S30.

Step 4: Use loop recursion to segment multiple adhesion particles. If a particle does not meet the particle segmentation principle, it cannot be segmented and exits directly. If the particle can be segmented, it is segmented into particle 1 and particle 2. Then determine the particle 1 and particle 2 in turn. Whether particle 2 satisfies the particle segmentation principle; if neither particle 1 nor particle 2 can continue to be segmented, exit; if there are particles that can continue to be segmented, continue segmentation; loop recursion until all particles are segmented, and the identification of coffee particles is obtained result.

The effect of segmentation of adherent particles is shown in Figure 5.

Step S70: Obtain the particle size evaluation result of the coffee particles based on the identification results of the coffee particles and the corresponding particle size and specific surface area statistical information.

Specifically, based on the above identification results of coffee particles, as well as the corresponding particle size and specific surface area statistical information, the particle size evaluation results of coffee particles are finally obtained, as shown in Figures 6 and 7. Figures 6 and 7 represent the coffee particle size. And the specific surface area evaluation results, where Figure 6 shows the particle size distribution histogram and particle size cumulative distribution histogram, and Figure 7 shows the specific surface area distribution histogram and the specific surface cumulative distribution histogram.

This invention first preprocesses the image of coffee particles, including grayscale and denoising, then uses Hough circle detection to find the center of the circular lamp, then converts the color image into a grayscale image, and then applies a Gaussian filter Denoising is performed. Next, the filtered grayscale image is converted into a binary image to distinguish the coffee particles from the background. Then the edge tracking algorithm of connectivity is used to identify the boundaries of the coffee particles and calculate the length and mean of each particle. , and then smooth the boundaries of the coffee particles through moving average filtering, and then extract the outer envelope of the particles to eliminate the interference of the particles in contact with each other on the experiment. Among the identified coffee particles, there may also be silver skin on the surface of the coffee beans. , the present invention defines the length and mean thresholds to determine whether it is silver skin to eliminate its interference; finally, in view of the situation where coffee particles may adhere, a particle segmentation algorithm based on the segmentation index is designed to segment the adhered particles. , if the segmented particles still meet certain shape and segmentation index conditions, retain them and continue segmenting until all particles are identified and segmented. Finally, according to the identification results of the coffee particles, as well as the corresponding particle size and specific surface area Statistical information is used to obtain the particle size evaluation results of coffee particles.

Further, as shown in Figure 8, based on the above automatic coffee particle size assessment method, the present invention also provides an automatic coffee particle size assessment system, wherein the coffee particle size automatic assessment system includes:

The coffee image preprocessing module 51 is used to obtain the original coffee particle size image, use the Hough circle detection algorithm to identify the central circular area of the original coffee particle size image, obtain the coffee particle size image, and process the coffee particle size image. Perform denoising and grayscale processing to obtain a grayscale image, and convert the grayscale image into a binary image;

The coffee particle profile boundary recognition module 52 is used to perform profile boundary recognition of coffee particles in the binary image using an edge tracking algorithm based on connectivity, extract the contours of all connected edges, and obtain the profile boundary;

The filtering and anti-aliasing module 53 is used to smooth the outline boundary and remove the aliasing of the outline boundary;

The particle outer envelope acquisition module 54 is used to obtain the particle convex hull according to the smoothed binary image, and perform processing according to the particle convex hull for each particle that has not obtained the outer envelope until all particles have obtained the outer envelope. network;

The coffee silver skin removal module 55 is used to confirm the particles belonging to silver skin according to the silver skin judgment conditions and remove the silver skin in the binary image;

The particle adhesion segmentation module 56 is used to segment the adhered coffee particles in the binary image using a particle segmentation algorithm based on the segmentation index to obtain the identification result of the coffee particles;

The particle size assessment module 57 is used to obtain the particle size assessment results of the coffee particles based on the identification results of the coffee particles and the corresponding particle size and specific surface area statistical information.

Further, as shown in FIG. 9 , based on the above-mentioned automatic coffee particle size assessment method and system, the present invention also provides a terminal. The terminal includes a processor 10 , a memory 20 and a display 30 . FIG. 9 only shows some components of the terminal, but it should be understood that implementation of all the components shown is not required, and more or less components may be implemented instead.

In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) equipped on the terminal. ) card, Flash Card, etc. Further, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed on the terminal, such as program codes for installing the terminal. The memory 20 can also be used to temporarily store data that has been output or is to be output. In one embodiment, the automatic coffee particle size assessment program 40 is stored in the memory 20, and the coffee particle size automatic assessment program 40 can be executed by the processor 10, thereby realizing the automatic coffee particle size assessment method in the present application.

In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run program codes or process data stored in the memory 20, for example Implement the automatic coffee particle size assessment method, etc.

In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, or the like. The display 30 is used to display information on the terminal and to display a visual user interface. The components 10-30 of the terminal communicate with each other via a system bus.

In one embodiment, when the processor 10 executes the coffee particle size automatic assessment program 40 in the memory 20, the steps of the coffee particle size automatic assessment method as described above are implemented.

The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores an automatic coffee particle size evaluation program. When the coffee particle size automatic evaluation program is executed by the processor, the coffee particle size automatic evaluation program is implemented as described above. The steps of automatic path assessment method.

In summary, the present invention provides an automatic coffee particle size assessment method and related equipment. The method includes: obtaining an original coffee particle size image, and using a Hough circle detection algorithm to identify the central circle of the original coffee particle size image. area, obtain a coffee particle size image, perform denoising and grayscale processing on the coffee particle size image to obtain a grayscale image, and convert the grayscale image into a binary image; the edge tracking algorithm based on connectivity Perform profile boundary recognition on the coffee particles in the binary image, extract the contours of all connected edges, and obtain the profile boundary; smooth the profile boundary to remove the jagged edges of the profile boundary; according to the smoothing process The particle convex hull is obtained from the binary image of Remove the silver skin in the binary image; use a particle segmentation algorithm based on segmentation index to segment the adhered coffee particles in the binary image to obtain the identification results of the coffee particles; according to the identification results of the coffee particles, and the corresponding particles diameter and specific surface area statistical information to obtain the particle size evaluation results of coffee particles. The present invention detects coffee particles based on image vision, which can greatly reduce detection costs and achieve accurate and efficient evaluation of the particle size and specific surface area of coffee particles.

It should be noted that, as used herein, the terms "include", "comprises" or any other variation thereof are intended to cover a non-exclusive inclusion, such that a process, method, article or terminal that includes a series of elements not only includes those elements, It also includes other elements not expressly listed or inherent in such process, method, article or terminal. Without further limitation, an element defined by the statement "comprises a..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes the element.

Of course, those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing relevant hardware (such as processors, controllers, etc.) through computer programs. The programs can be stored in a computer. In a computer-readable computer-readable storage medium, when executed, the program may include the processes of the above method embodiments. The computer-readable storage medium may be a memory, a magnetic disk, an optical disk, etc.

It should be understood that the application of the present invention is not limited to the above examples. Those of ordinary skill in the art can make improvements or changes based on the above descriptions. All these improvements and changes should fall within the protection scope of the appended claims of the present invention.

Claims (10)

1. An automatic evaluation method of coffee particle size, characterized in that the automatic evaluation method of coffee particle size comprises:

acquiring an original coffee particle size image, identifying a central circular area of the original coffee particle size image by adopting a Hough circle detection algorithm to obtain a coffee particle size image, carrying out denoising treatment and gray scale treatment on the coffee particle size image to obtain a gray scale image, and converting the gray scale image into a binary image;

carrying out contour boundary identification on coffee particles in the binary image based on a connectivity-based edge tracking algorithm, extracting the contours of all connected edges, and obtaining contour boundaries;

smoothing the profile boundary to remove saw teeth of the profile boundary;

obtaining a particle convex hull according to the smoothed binary image, and processing each particle which does not obtain an outer envelope according to the particle convex hull until all the particles obtain the outer envelope;

confirming particles belonging to the silver skin according to the silver skin judging conditions, and removing the silver skin in the binary image;

dividing the adhered coffee particles in the binary image by a particle dividing algorithm based on the dividing index to obtain a recognition result of the coffee particles;

and obtaining a particle size evaluation result of the coffee particles according to the identification result of the coffee particles and the corresponding particle size and specific surface area statistical information.

2. The method for automatically evaluating the particle size of coffee according to claim 1, wherein the steps of obtaining an original particle size image of coffee, identifying a central circular area of the original particle size image of coffee by using hough circle detection algorithm, obtaining a particle size image of coffee, performing denoising and gray scale processing on the particle size image of coffee to obtain a gray scale image, and converting the gray scale image into a binary image, comprise:

acquiring an original coffee particle size image acquired by a CMOS camera, and identifying a central circular area with brightness higher than preset brightness in the original coffee particle size image by adopting a Hough circle detection algorithm to acquire a coffee particle size image;

denoising the coffee particle size image by using a Gaussian filter, wherein if the two-dimensional Gaussian function is:

wherein sigma is the standard deviation of a Gaussian function, x and y represent the abscissa and the ordinate of a pixel point in an original coffee particle size image, and a smooth image K (x, y) is obtained by convolving rows and columns of the original coffee particle size image f (x, y) with the Gaussian function G (x, y);

carrying out gray level processing on the denoised smooth image K (x, y) by adopting a weighted average method to obtain a gray level image, wherein the formula is as follows:

I＝αR+βG+γB；

wherein R, G, B is 3 components of a certain pixel of the original coffee particle size image, alpha, beta and gamma are intensity coefficients, and the alpha+beta+gamma=1 is satisfied, and I represents a gray level image;

and converting the gray level image into a binary image after binarization treatment, wherein the binary image is used for representing coffee particles and background colors.

3. The method for automatically evaluating the particle size of coffee according to claim 2, wherein the connectivity-based edge tracking algorithm performs contour boundary recognition on coffee particles in the binary image, extracts the contour of all connected edges, and obtains the contour boundary, and specifically comprises:

selecting any one white pixel point in the binary image as a starting point;

starting from the starting point, scanning surrounding 8 pixel points in a clockwise or anticlockwise direction in sequence, searching a black pixel point as a next point, marking the black pixel point as a current point if the black pixel point is found, continuing the next scanning, and ending tracking if the black pixel point is not found;

after the next point is found, marking a connecting line between the current point and the next point as an edge, taking the next point as the current point, and continuing to perform the next scanning;

ending the tracking when returning to the starting point, and returning all marked edge points as one contour;

if a plurality of connected contours exist, repeating scanning from an unaccessed starting point until all contours are extracted to obtain a contour boundary, and recording the pixel length of the ith grain boundary after the contour boundary is obtained _i And the mean of the gray values of all pixels of the ith particle _i 。

4. The method according to claim 3, wherein the smoothing the profile boundary to remove saw teeth of the profile boundary comprises:

smoothing the contour boundary of coffee particles by moving average filtering, and obtaining a moving average filtered output signal during the process of averagingMake the output signal +.>Error between the measured values x (i) in one sampling period>The sum of squares is minimal, where i=1, 2, … …, N, then there are:

N ^* ＝argmin(W)；

obtaining an extremum, and obtaining:

where N represents the number of points used in moving average filtering, W represents an objective function that requires minimization, N ^* Representing a point or set that minimizes the objective function W;

after the profile array is spliced, moving average filtering is carried out, if the initial profile boundary has n data points, the point set is { (x) ₁ ,y ₁ )、(x ₂ ,y ₂ )、……、(x _n-1 ,y _n-1 )、(x ₁ ,y ₁ ) "written as:

taking the sub-matrix of the rear (n-1) row of the matrix A as follows:

splicing the matrix A and the submatrix B to obtain:

moving average filtering is respectively carried out on column vectors of the matrix C:

when it is obtainedAfter the calculation result of one point, the subsequent filtering result is calculated by using the preamble result:

wherein p= (N-1)/2, q=p+1;

the filtered matrix is noted as:

taking a matrixThe n rows of submatrices D of (a) are the filtering results of the profile boundary point set:

wherein i=n/2+1,representing the average value of the measured values x (i);

and sequentially carrying out filtering treatment on each profile boundary which is not subjected to filtering and antialiasing until all the profile boundaries are smooth.

5. The method according to claim 4, wherein the step of obtaining a convex hull of particles from the smoothed binary image, and for each particle not having obtained an envelope, processing according to the convex hull of particles until all particles have obtained an envelope, comprises:

if P ₁ (x ₁ ,y ₁ )、P ₂ (x ₂ ,y ₂ )、……、P _n (x _n ,y _n ) For smoothing the set of n points on the boundary of the back profile, wherein x is arranged from small to large, if x is equal, y is arranged from small to large, and the point P with the smallest abscissa is selected ₁ If the abscissas are the same, the minimum point of the abscissas is selected as the starting point, P ₁ For convex hull vertices of the set, then:

wherein i is E (2, n)]If (if)(n-2) results are all numbered, then P ₂ As a point on the convex hull, otherwise, P ₂ Not points on the convex hull;

if P ₂ Is the point on the convex hull, then:

wherein j is E (3, n)]If (if)(n-3) results are all numbered, then P ₃ As a point on the convex hull, otherwise, P ₃ Not points on convex hull, P ₄ 、P ₅ 、……、P _n Sequentially testing in the same mode until the traversal is finished, and finding out all points conforming to the convex hull conditions;

for each particle not obtaining the outer envelope, sequentially processing until all particles obtain the outer envelope;

for any polygon, the coordinates of each vertex are A ₁ (x ₁ ,y ₁ ),A ₂ (x ₂ ,y ₂ )，……，A _n (x _n ,y _n ) The perimeter and area of the polygon are:

wherein x is _n+1 ＝x ₁ ，y _n+1 ＝y ₁ ；

Respectively calculating the perimeter and the area of the particle profile, and respectively marking as P ₁ 、S ₁ The method comprises the steps of carrying out a first treatment on the surface of the Calculating the perimeter and the area of the convex hulls of the particles, respectively marked as P ₂ 、S ₂ ；

Defining the ratio of the actual area of the particles to the area of the convex hull as a shape factor, and recording as

6. The method according to claim 5, wherein the step of determining particles belonging to the silver skin according to the silver skin determination condition, and removing the silver skin from the binary image, comprises:

setting a granular pixel length threshold alpha ₁ 、α ₂ And a mean threshold value beta for the particle pixel length ₁ 、β ₂ ；

The silver skin judging conditions include:

silver skin judging condition one: ith grain boundary pixel length _i >α ₁ And mean value mean _i >β ₁ ；

Silver skin judging condition II: ith grain boundary pixel length _i >α ₂ And mean value mean _i >β ₂ ；

If the ith particle satisfies either one of the first silver skin judging condition and the second silver skin judging condition, the ith particle is considered to be silver skin, and the silver skin is removed.

7. The method for automatically evaluating the particle size of coffee according to claim 6, wherein the particle segmentation algorithm based on the segmentation index segments the adhered coffee particles in the binary image to obtain the recognition result of the coffee particles, and specifically comprises:

if Q ₁ ＝(x ₁ ,y ₁ )、Q ₂ ＝(x ₂ ,y ₂ )、……、Q _n ＝(x _n ,y _n ) N points on the profile boundary, and x is arranged from small to large, and if x is equal, y is arranged from small to large, where Q _i ＝(x _i ,y _i )、Q _j ＝(x _j ,y _j ) Is formed by two points (j>i) If European distance D ₁ For the straight line distance between two points, then:

if the perimeter distance D ₂ Is Q _i The point reaches Q along the contour boundary _j Shortest distance of points, perimeter distance D ₂ Approximate the difference in sequence numbers:

obtaining Euclidean distance D of any two points on profile boundary ₁ And circumference distance D ₂ The method comprises the steps of carrying out a first treatment on the surface of the The ratio is:

ratio＝D ₁ /D ₂ ；

two points exist on the outline boundary of the particle to minimize the value of the ratio, the two points are taken as key points, and the value of the minimum ratio is taken as a segmentation index and is recorded as min { ratio };

setting a particle shape factor threshold gamma, a segmentation index threshold epsilon, and defining a particle segmentation principle comprises:

particle segmentation principle one: shape factor shape of particles _factor >Gamma and division index min { ratio }<ε；

Particle segmentation principle two: a key point sequence number (j-i) >3 meeting the first particle segmentation principle;

if the particles meet the first particle segmentation principle and the second particle segmentation principle at the same time, the particles are segmented into particles 1 and particles 2, otherwise, the particles are not segmented;

smoothing the two grain divisions, if the pixel length gra of the grain 1 boundary ₁ >4, smoothing the division of the particles 1; if the pixel length gra of the grain 2 boundary ₁ >4, smoothing the division of the particles 2;

dividing a plurality of particles by using cyclic recursion, if one particle does not meet the particle dividing principle, the particles cannot be divided and directly exit; if the particles can be segmented, dividing the particles into particles 1 and particles 2, and sequentially judging whether the particles 1 and the particles 2 meet a particle segmentation principle; if both particles 1 and 2 cannot be continuously segmented, exiting; if the particles exist, the division can be continued, and the division is continued; and (5) circulating recursion until all the particles are completely segmented, and obtaining a coffee particle identification result.

8. An automatic coffee particle size assessment system, characterized in that the automatic coffee particle size assessment system comprises:

the coffee image preprocessing module is used for acquiring an original coffee particle size image, identifying a central circular area of the original coffee particle size image by adopting a Hough circle detection algorithm to obtain a coffee particle size image, carrying out denoising treatment and gray level treatment on the coffee particle size image to obtain a gray level image, and converting the gray level image into a binary image;

the coffee particle profile boundary recognition module is used for performing profile boundary recognition on coffee particles in the binary image based on a connectivity edge tracking algorithm, extracting the profiles of all communicated edges and obtaining a profile boundary;

the filtering and antialiasing module is used for performing smoothing treatment on the profile boundary and removing the sawteeth of the profile boundary;

the particle outer envelope acquisition module is used for obtaining a particle convex hull according to the smoothed binary image, and processing each particle without the outer envelope according to the particle convex hull until all the particles obtain the outer envelope;

the coffee silver skin removing module is used for confirming particles belonging to silver skin according to silver skin judging conditions and removing the silver skin in the binary image;

the particle adhesion segmentation module is used for segmenting the adhered coffee particles in the binary image based on a particle segmentation algorithm of a segmentation index to obtain a recognition result of the coffee particles;

and the particle size evaluation module is used for obtaining the particle size evaluation result of the coffee particles according to the recognition result of the coffee particles and the corresponding particle size and specific surface area statistical information.

9. A terminal, the terminal comprising: memory, a processor and a coffee particle size automatic assessment program stored on the memory and executable on the processor, which when executed by the processor, implements the steps of the coffee particle size automatic assessment method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a coffee particle size automatic evaluation program which, when executed by a processor, implements the steps of the coffee particle size automatic evaluation method according to any one of claims 1 to 7.