METHOD OF IDENTIFYING ABNORMAL MOTION OBJECT IN COMPRESSED IMAGE USING MOTION VECTOR TRAJECTORY AND PATTERN

摘要

The present invention relates to a technique of effectively identifying an abnormal motion object, that is, an object that exhibits unique behavior which is different from other objects generally in a compressed image such as H.264 AVC and H.265 HEVC. In particular, the present invention relates to a technique of extracting a moving object region using a motion vector obtained by parsing a compressed image and effectively identifying an abnormal motion object on the basis of a motion vector trajectory pattern, rather than identifying the abnormal motion object by recognizing the existence of objects through a complex image analysis as in prior art, for example, through a compressed image generated by a CCTV camera, and observing the behavior of the objects. In particular, the present invention relates to a technique of improving the accuracy of identification of an abnormal motion object by constructing a training dataset with a motion vector trajectory pattern obtained from a compressed image and learning a neural network to be utilized for identifying the abnormal motion object.