An adaptive video object behavior track analysis method based on a convolutional neural network

Abstract

The invention discloses an adaptive video object behavior track analysis method based on a convolutional neural network, and belongs to the technical field of image engineering and machine vision. Fora monitoring video in a real scene, a kernel correlation filtering algorithm is adopted to effectively detect and track a video object, so that a track graph of a directed curve is generated, a convolutional neural network is utilized based on deep learning to adaptively train and learn a normal behavior track in the current scene, and finally suspicious behaviors are distinguished; According tothe invention, deep learning is applied to behavior analysis of the video object; the normal behavior track in the scene can be learned in a self-adaptive manner; The method has the advantages that manual definition is not needed, so that the suspicious behavior judgment reliability is guaranteed, the use background can be continuously expanded in a real scene, the detection tracking part utilizesthe characteristics of a cyclic matrix, the deep learning input is a two-dimensional track graph, the space-time characteristics of a video are reserved, and the real-time performance is guaranteed to the maximum extent.

Description

technical field [0001] The invention belongs to the technical fields of image engineering and machine vision, and in particular relates to a behavior analysis method of video objects. Background technique [0002] With the advent of the era of artificial intelligence, intelligent video surveillance systems have become an important part. Under the surveillance camera, for the video objects in the real scene, it can automatically detect and track in real time, and conduct effective behavior analysis and understanding to realize intelligent video surveillance in the true sense. It needs to be integrated with mathematics, computer, electronics, control, psychology and other disciplines, and can play a key role in many fields such as national security, traffic management, and social services. This topic has high theoretical research value and broad practical application prospects. [0003] Detection and tracking of moving targets: From the perspective of research objects, it ca...

AI Technical Summary

Problems solved by technology

The difficulty of detection and tracking still lies in complex background situations such as occlusion and scale changes

Images