Automatic Recognition And Detection Of Moving Objectives

Automatic recognition and detection of moving objectives is a heat and difficult subject in the field of computer vision, image processing and pattern recognition. A quick and accurate motion detecting has great practical utility in the field of intellectual monitoring, pedestrians counting system, video compression and traffic monitoring systems ect.Recognition and detection of moving objectives of dynamic image is a important part of image processing. This subject is made up of detection of moving object, segmentation and feature extraction of object, in which central issue of image processing and pattern identification is involved. The paper makes a study of problem of moving objectives detecting in real-time scene. Firstly, a kind of solidity and exact algorithm of recognition and detection of moving objectives has been proposed. An experimental system is built , which is used for identifying and detecting more moving objectives. At last it accomplishes the task of image capturing, saving and recognition and counting of moving objectives.Concerning the algorithm, a new kind of modified algorithm aimming at moving objectives detecting has been proposed in the paper. When there is a lot of small noise and pseudo-objectives in current frame, this algorithm could restrain the change of lighting and bring increased accuracy to recognition. The technique of image filtering, image segmentation, morphology processing, shadow suppression is used in image processing. The part in motion is extracted almostly. Then, a modified method of pixel marker is proposed. The method gives mark to each object target in order and the number of moving targets would been counted exactly. The feature of motion area, that is of different mark in labeled image, is extracted . So that the objects are detected one by one, The undesirable one was combed out . Finally, we could achieve the statistics of the number of targets really and truly.