WO1997018666A1

WO1997018666A1 - Screening method

Info

Publication number: WO1997018666A1
Application number: PCT/SE1996/001451
Authority: WO
Inventors: Björn KRUSE
Original assignee: Forskarpatent I Linköping Ab
Priority date: 1995-11-13
Filing date: 1996-11-12
Publication date: 1997-05-22
Also published as: SE9504016L; AU7659496A; SE9504016D0

Abstract

Electronic, digital screening of pictures, in particular for printing, where different types of screens are mixed for one and the same picture. In picture areas with a detailed structure a fine screen is used while for larger areas with a less detailed structure in grey or colour scales a more coarse screen is used. Coarser screens may further be used for the parts of the picture that are to be very light (for instance skin tones and shiny parts). By allowing the screens to be integer number multiples of each other automatically soft transitions are obtained and some picture points will be parts in both screens.

Description

Screening method

Background

Picture reproduction in print is in most cases done with a very limited number of colours. For the reproduction of black and white pictures only one colour is necessary, normally black, while reproduction of colour picture requires at least three printing colours.

In order to resemble the continuous tone variation of the original screens are used. The object of the screen is to distribute the colour over the surface, surface modulation, so that the resulting impression of the screened picture coincides with that of the original. If the tone value 50% is desired the area is in principle to be divided so that half of the area is covered with colour and half of the area is bare (without colour). This dividing of the area can be carried out in a number of different ways. A usual method of digital screening is to imitate the effect of differently large screen points that one achieves at optical screening. Another way is to emulate the light sensitive particles in a film. In the latter case the result receives a random character. In the former case the result is regularly arranged as a lattice.

At the printing of colour pictures the original picture is separated in its colour components, most frequently cyan, magenta and yellow that are each screened separately. Because of the risk of moire patterns different techniques are used to modify the conventional screens, for instance by the use of different screen angles of the separations . If non-periodic screens are used the moire problem is minimised already from the start .

Different types of screens have different properties in combination with the character of the picture and the media on which it is printed. Generally speaking one can say that it is more difficult to use screens with small screen points. A fine screen with small points, however, provides the possibility of increased reproduction of details. It is therefore desirable to use a fine screen where there are fine details in the picture while a coarser screen is to be preferred for the reproduction of even tone surfaces without details . This patent application concerns a method to achieve a mixed screen that meets these desired objects but is not limited on this. Description of the method

The method depends on the original picture being divided into segments for which different screening methods are suitable. The separating into segments can be done based on local richness of detail, tone value or some other property of the picture or the media. Each point in the picture area belongs to one of the different segments. In the transition zones between two segments a point may partly belong to two or several segments. For instance the local variation may be given a threshold so that a segment is constituted by detail, in the picture with great variation, S₂, while the second segment has a lower local variation, S₁. If our starting picture is P and N is the local environment the following statement can be made :

where t₂ is a threshold that rules the size of the variation. Only points with a greater local variation belong to. S₂. The segment S₁ is in this case constituted by the remaining picture points, that is those that do not belong to

. The variation may be constituted by the variation measure known from statistics but may also be calculated in other ways It should be observed that the segments are not necessarily hanging together. How fragmented they are is ruled by the character of the original picture . In a more complicated case with three segments S₁ , S₂ and S₃, the definition could be the following.

S₁ remaining points In this case the picture is separated into three segments. One, S ₃, where there is a large richness of detail, one, S₂, where the richness of detail is less but where the tone value is higher than the threshold value t₂ and finally S₁ constituted by the remaining points where we have a low tone value and little variation. The above examplifies two possible segementation grounds. The number of segments may of course be larger. In the following we allow n to define the number of segments S₀, S₁ and

S_n. In the examples above the segments have been binary, that is each point in the picture area has belonged to only one segment, in this case:

Furthermore

that is each point (x,y) in the area belongs to one and only one segment. If, however, 0.0 < S₁(x,y) < 1 .0 the point (x,y) lies in a transition zone between the two segments. The value of S_i (x,y) may be associated to how close the point is to the inner of the segment. A large value close to 1.0, means that the point is closer to the inner of the segment while a lower value means that it is further away. The border zone may for instance be made by low pass filtration of the segment in question . At the transition between two segments values between 0.0 and 1.0 result defining the degree of belongingness .

Now let R₁, R₂... R_n, be n number of screens belonging to respective segments. The screens may explicitly be defined as matrixes or functions of (x,y) so that a screen value between 0.0 and 1.0 can be calculated for each picture point. The screen may for instance be a conventional screen, a stocastic screen or a screen that has been made in some other optional way. By the separating into segments we can make new screens R by

R is the starting point for an optimising (minimising) of the difference between the original picture and the screened version of the original picture. The screened picture is obtained by comparing the picture point value with the corresponding screen point value (threshold screening) . Let RP denote the screened picture. We can write :

Furthermore allow D(x,y) to be the difference between the original and screened version added over a surrounding N:

where N is a surrounding of the point (x,y) If the original picture were binary D would be≡ 0 at optimum. If P however is a picture with half tones D will never be zero everywhere. Our requirement for optimum is instead that |D(x,y)| is small in each point. A more simple generalisation of the difference matrix D is a weighted sum over the points of the surrounding:

where

Previously (without this generalisation) all coefficients have one and the same value. If D_Ν≡ 0 the method is clear. Otherwise the screen is modified successively so that |D_Ν| is minimised. This is done in the following way by making a new screen matrix R ':

The picture is screened with the new screen R' . If the error or deviation is diminished the screen R may be updated by putting R = R' and the process is repeated yet another time. The process may be continued as long as |D_N| decreases. When this is no longer the case a minimum point has been achieved and the process is ended. In order to screen a colour picture the method is used on the separations of the picture separately . An alternative method is to include all colours in the optimising process. By calculation in each iteration of the optimisation in which separation that an altering reduces the error most and thereafter chose that for updating the process may better adapt itself to the variation of the picture.

Point enlarging is an effect at the producing of films, plates and in printing by the interacting of the colour with the paper. When illuminated an effect is added depending on optical spreading effects in the paper substrate. The first effect is normally called mechanical point enlarging while the latter is called optical point enlarging . A simple correction of the point enlarging can be made in the tone value of the picture, as is common today, however the point enlarging is dependent both on the shape of the screen and the size of the screen points. A compensation that in this meaning takes detailed regard to shape and size can only be executed under the optimising phase above . It is not until the picture is screened that the real look of the actual points are known.

A simple generalisation of the method takes during the optimising sequence regard to the above described local point enlarging . At the calculation of the difference function the expression D_N is changed to: where

In this expression G(RP) is a generally unlinear function over an environment of each point in RP. The function G is such that is compensates for the point enlarging actually obtained. The function has a different form depending on actual printing media and printing press. For instance the influence of the physical point enlarging on RP can be described as:

RP' is now no longer binary but shows an extended zone outside each screen point, that depending on the coefficients c_{ζ η} may have different extension . The influence of the optical point enlarging on RP' may for instance be described as :

The coefficients d_{ζ η} are ruled by the optical spreading properties of the paper. The function G may also take in regard the remaining frequency response of the screen function. A simple generalisation of this is to allow G also to be dependent on the specific screen function.

In order to illustrate the invention a picture is enclosed which has been screened in accordance with the invention. Only two screens have been used, which are of the same type but with a doubled number of points for the finer screen i relation to the more coarse one. Since the screens are integer number multiples of each other automatically a soft transition is obtained between the filters and some picture points will be parts of both screens.

Claims

1 . Electronic, in particular digital screening of pictures, in particular for printing,

characterised in that different types of screens are mixed for one and the same picture, the choice of screens being dependent on what is to be depicted.

2. Screening according to claim 1, characterised in that in picture parts with a detailed structure a fine screen is used while for larger areas with less detailed structure in grey or colour scales a more coarse screen i used.

3 . Screening according to claim 1 , characterised in that the transition between different screens is successive, by for instance bringing a digital picture signal, that is used as a masque, to pass through a low pass filter .

4. Screening according to any of the preceding claims, characterised in that the used screens are of the same type and in particular multiples of each other.

5. Screening according to any of the preceding claims, characterised in that after the screening is carried out, a comparison is made of the screened picture and the original picture, the differences in appearance of the screened picture being used locally to correct the screened picture, for instance the size of the points, which method may be iteratively repeated in order successively to give an increasing coincidence in perception between the original picture and the screened picture.

6. Screening according to claim 5, characterised in that simultaneously a correction for optical point enlarging is made.

7. Screening according to any of the preceding claims, characterised in that coarser screens are used for the parts of the picture that are to be very light (for instance skin tones and shiny parts).