+

WO2018131729A1 - Procédé et système de détection d'un objet mobile dans une image à l'aide d'une seule caméra - Google Patents

Procédé et système de détection d'un objet mobile dans une image à l'aide d'une seule caméra Download PDF

Info

Publication number
WO2018131729A1
WO2018131729A1 PCT/KR2017/000359 KR2017000359W WO2018131729A1 WO 2018131729 A1 WO2018131729 A1 WO 2018131729A1 KR 2017000359 W KR2017000359 W KR 2017000359W WO 2018131729 A1 WO2018131729 A1 WO 2018131729A1
Authority
WO
WIPO (PCT)
Prior art keywords
divided
image
optical flow
region
object detection
Prior art date
Application number
PCT/KR2017/000359
Other languages
English (en)
Korean (ko)
Inventor
김정호
최병호
황영배
장성준
Original Assignee
전자부품연구원
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 전자부품연구원 filed Critical 전자부품연구원
Publication of WO2018131729A1 publication Critical patent/WO2018131729A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/20—Analysis of motion
    • G06T7/215—Motion-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/90—Determination of colour characteristics
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/80—Camera processing pipelines; Components thereof

Definitions

  • the present invention relates to an object detection method, and more particularly, to a method and system for determining and detecting a moving object from an image sequence obtained by a single camera.
  • a stereo camera or a single camera can be used to calculate the movement of the camera and perform detection therefrom.
  • the technique of estimating camera movement has a problem in that accuracy decreases when there are many moving objects.
  • the present invention has been made to solve the above problems, and an object of the present invention is to provide an object detection method and system that can improve the accuracy in detecting a moving object based on a single camera.
  • an object detecting method includes: generating an optical flow image from an input image sequence; Extracting divided regions from an optical flow image; And generating a motion region from the extracted divided regions.
  • the motion regions may be generated by integrating or maintaining the divided regions through color comparison between the divided regions and neighboring divided regions.
  • the color difference values between the divided areas and the difference value in the boundary line may be compared to integrate or maintain the divided areas.
  • the moving region may be generated by maintaining or removing the divided region based on the size and position of the divided region in the optical flow image.
  • the size of the divided region may be the height of the divided region, and the position of the divided region may be the y-coordinate at the bottom of the divided region.
  • the input image sequence may be generated using a single camera.
  • the object detection system for generating an input image sequence; And a processor configured to generate an optical flow image from the input image sequence generated by the camera, extract the divided regions from the optical flow image, and generate a motion region from the extracted divided regions.
  • object detection performance may be improved by using an optical flow image, an image segmentation technique, and an object detection using an approximate size of a moving object using a single camera. Will be.
  • embodiments of the present invention can be applied to an intelligent driverless vehicle, and can be applied to a technology capable of recognizing such a situation.
  • FIG. 2 is a diagram illustrating an input image
  • FIG. 3 is a diagram illustrating an optical flow image of FIG. 2;
  • FIG. 4 is a diagram illustrating a result of extracting divided regions from an optical flow image
  • 5 to 8 are views provided for further explanation of a post-processing process for motion regions
  • 9 to 12 are views illustrating results of detecting a moving object from a single camera image by a method according to an embodiment of the present invention.
  • FIG. 13 is a block diagram of an object detection system according to another embodiment of the present invention.
  • the object detection method according to an embodiment of the present invention accurately detects moving objects in an image sequence generated by using a single camera.
  • the object detection method by using the image sequence obtained by a single camera to calculate the amount of movement of the pixels in the optical flow (moving by splitting the image through the size and direction of the flow) Detect objects and improve the performance of object detection through the approximate size of moving objects.
  • the object detection method by comparing the input image (current image) and the previous image in the image sequence generated using a single camera, Optical Flow ) Generates an image (S110).
  • FIG. 2 illustrates an input image
  • FIG. 3 illustrates an optical flow image of FIG. 2.
  • segmented regions are extracted by dividing adjacent pixels having the same colors that are determined to have a motion in the optical flow image generated in S110 (S120). 4 illustrates the results of extracting the divided regions from the optical flow image.
  • step S130 motion regions are generated from the divided regions obtained in operation S120 (S130).
  • the movement area generation in step S130 is performed by grouping the divided areas obtained in step S120 or removing / excluding a portion.
  • each partition is configured as a node in the graph, and neighboring partitions are connected as edges to compare two colors or to separate the partitions through color comparison.
  • Equation 1 the maximum color value in the partition is calculated by Equation 1 below.
  • Equation 2 The minimum color difference value at the boundary line between two neighboring partitions is calculated by Equation 2 below.
  • Equation 3 if the maximum color difference value, which is the difference value between the maximum color values inside the two partitions, is smaller than the minimum color difference value at the boundary line, the two partitions remain divided and vice versa. In this case, two partitions are combined and grouped into one partition.
  • k represents a constant value and ⁇ represents the number of pixels belonging to the divided region.
  • the motion regions determined as not the actual moving objects are removed by inferring whether the size of the movement regions corresponds to the size of the actual moving objects.
  • Equation 4 The correlation between the y coordinate and h is calculated from the linear relationship shown in Equation 4 below.
  • a linear function can be calculated by projecting an image at an interval of about 10 cm from an actual height of 1 m to 3 m (see FIG. 6), calculating a matrix A, calculating eigenvectors and eigenvalues from SVD (Singular Value Decomposition)
  • SVD Single Value Decomposition
  • h min and h max corresponding to y coordinates can be calculated.
  • the partition is determined to be not a moving object and removed.
  • FIG. 8 a result of removing regions which do not correspond to actual moving objects among the movement regions shown in FIG. 4 is shown in FIG. 8.
  • 9 to 12 illustrate the results of detecting a moving object from a single camera image by the method according to an embodiment of the present invention. It can be seen that the robust moving object can be detected by overcoming the limitation of estimating motion without distance information, which is the limitation of a single camera.
  • the object detection system includes a camera 110, an image processor 120, and an output unit 130 as shown in FIG. 13.
  • the camera 110 generates an image sequence with a single camera system, and the image processor 120 detects a moving object from a single camera image through the algorithm shown in FIG. 1.
  • the output unit 130 may be various means for outputting / saving a detection result of a moving object, such as a display, an interface, a memory, and the like.
  • the technical idea of the present invention can be applied to a computer-readable recording medium containing a computer program for performing the functions of the apparatus and method according to the present embodiment.
  • the technical idea according to various embodiments of the present disclosure may be implemented in the form of computer readable codes recorded on a computer readable recording medium.
  • the computer-readable recording medium can be any data storage device that can be read by a computer and can store data.
  • the computer-readable recording medium may be a ROM, a RAM, a CD-ROM, a magnetic tape, a floppy disk, an optical disk, a hard disk drive, or the like.
  • the computer-readable code or program stored in the computer-readable recording medium may be transmitted through a network connected between the computers.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Signal Processing (AREA)
  • Image Analysis (AREA)

Abstract

L'invention concerne un procédé et un système de détection d'un objet mobile dans une image à l'aide d'une seule caméra. Un procédé de détection d'objet, selon un mode de réalisation de la présente invention, génère une image de flux optique à partir d'une séquence d'image d'entrée, extrait des régions segmentées de l'image de flux optique, et génère une région de mouvement à partir des régions segmentées extraites. Ainsi, la présente invention peut améliorer les performances de détection d'objet, et peut être appliquée à une technologie de sensibilisation au contexte, telle qu'un véhicule sans pilote intelligent et analogues.
PCT/KR2017/000359 2017-01-11 2017-01-11 Procédé et système de détection d'un objet mobile dans une image à l'aide d'une seule caméra WO2018131729A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR1020170003980A KR102336284B1 (ko) 2017-01-11 2017-01-11 단일 카메라를 이용한 영상에서 움직이는 객체 검출 방법 및 시스템
KR10-2017-0003980 2017-01-11

Publications (1)

Publication Number Publication Date
WO2018131729A1 true WO2018131729A1 (fr) 2018-07-19

Family

ID=62840539

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2017/000359 WO2018131729A1 (fr) 2017-01-11 2017-01-11 Procédé et système de détection d'un objet mobile dans une image à l'aide d'une seule caméra

Country Status (2)

Country Link
KR (1) KR102336284B1 (fr)
WO (1) WO2018131729A1 (fr)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110007107B (zh) * 2019-04-02 2021-02-09 上海交通大学 一种集成不同焦距摄像头的光流传感器
WO2020242179A1 (fr) * 2019-05-29 2020-12-03 (주) 애니펜 Procédé, système et support d'enregistrement lisible par ordinateur non transitoire pour fournir du contenu

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20110118376A (ko) * 2010-04-23 2011-10-31 동명대학교산학협력단 광류를 이용한 방범용 차량 감지 시스템
KR20130060274A (ko) * 2010-08-02 2013-06-07 페킹 유니버시티 효과적인 비디오 분류 및 검색을 위한 대표 모션 플로우 추출
KR20130075636A (ko) * 2011-12-27 2013-07-05 중앙대학교 산학협력단 배경 합성을 위한 자동 객체 분할 장치 및 방법
KR20150089677A (ko) * 2014-01-28 2015-08-05 엘지이노텍 주식회사 카메라 시스템, 캘리브레이션 장치 및 캘리브레이션 방법
KR20150113751A (ko) * 2014-03-31 2015-10-08 (주)트라이큐빅스 휴대용 카메라를 이용한 3차원 얼굴 모델 획득 방법 및 장치

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2004088599A (ja) * 2002-08-28 2004-03-18 Toshiba Corp 画像監視装置と画像監視方法
JP2008226261A (ja) * 2008-04-07 2008-09-25 Toshiba Corp オブジェクト検出方法
KR101141936B1 (ko) * 2010-10-29 2012-05-07 동명대학교산학협력단 광류장 기반의 손 이미지 영역 검출 방법 및 그 검출 방법에 의한 손동작 인식 방법

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR20110118376A (ko) * 2010-04-23 2011-10-31 동명대학교산학협력단 광류를 이용한 방범용 차량 감지 시스템
KR20130060274A (ko) * 2010-08-02 2013-06-07 페킹 유니버시티 효과적인 비디오 분류 및 검색을 위한 대표 모션 플로우 추출
KR20130075636A (ko) * 2011-12-27 2013-07-05 중앙대학교 산학협력단 배경 합성을 위한 자동 객체 분할 장치 및 방법
KR20150089677A (ko) * 2014-01-28 2015-08-05 엘지이노텍 주식회사 카메라 시스템, 캘리브레이션 장치 및 캘리브레이션 방법
KR20150113751A (ko) * 2014-03-31 2015-10-08 (주)트라이큐빅스 휴대용 카메라를 이용한 3차원 얼굴 모델 획득 방법 및 장치

Also Published As

Publication number Publication date
KR20180082739A (ko) 2018-07-19
KR102336284B1 (ko) 2021-12-08

Similar Documents

Publication Publication Date Title
WO2022050473A1 (fr) Appareil et procédé d'estimation de pose de caméra
US11928800B2 (en) Image coordinate system transformation method and apparatus, device, and storage medium
CN106952269B (zh) 近邻可逆的视频前景物体序列检测分割方法及系统
WO2015126031A1 (fr) Procédé de comptage de personnes et dispositif associé
WO2019169884A1 (fr) Procédé et dispositif de détection de saillance d'image fondés sur des informations de profondeur
WO2013048160A1 (fr) Procédé de reconnaissance de visage, appareil et support d'enregistrement lisible par ordinateur pour exécuter le procédé
CN109033972A (zh) 一种目标检测方法、装置、设备及存储介质
CN111382637B (zh) 行人检测跟踪方法、装置、终端设备及介质
CN106997478B (zh) 基于显著中心先验的rgb-d图像显著目标检测方法
KR101963404B1 (ko) 2-단계 최적화 딥 러닝 방법, 이를 실행시키기 위한 프로그램을 기록한 컴퓨터 판독 가능한 기록매체 및 딥 러닝 시스템
WO2017150899A9 (fr) Procédé de ré-identification d'objet pour un suivi d'objets multiples global
WO2019147024A1 (fr) Procédé de détection d'objet à l'aide de deux caméras aux distances focales différentes, et appareil associé
CN105740751A (zh) 一种目标检测与识别的方法和系统
CN114072839A (zh) 单目静止相机视频中的分层运动表示与提取
CN107610177A (zh) 一种同步定位与地图构建中确定特征点的方法和设备
KR20230166840A (ko) 인공지능을 이용한 객체 이동 경로 확인 방법
WO2023050810A1 (fr) Procédé et appareil de détection de cible, dispositif électronique, support d'enregistrement et produit programme d'ordinateur
WO2014185691A1 (fr) Appareil et procédé d'extraction d'image à filigrane élevé à partir d'images photographiées en continu
CN117593548A (zh) 基于加权注意力机制去除动态特征点的视觉slam方法
WO2018131729A1 (fr) Procédé et système de détection d'un objet mobile dans une image à l'aide d'une seule caméra
CN113129249B (zh) 基于深度视频的空间平面检测方法及其系统和电子设备
WO2016104842A1 (fr) Système de reconnaissance d'objet et procédé de prise en compte de distorsion de caméra
Bikmullina et al. Stand for development of tasks of detection and recognition of objects on image
JP5027201B2 (ja) テロップ文字領域検出方法，テロップ文字領域検出装置およびテロップ文字領域検出プログラム
CN114399729A (zh) 监控对象移动识别方法、系统、终端及存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17890997

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17890997

Country of ref document: EP

Kind code of ref document: A1

点击 这是indexloc提供的php浏览器服务，不要输入任何密码和下载