Moving target detection method in slowly-changing moving background

Abstract

The invention relates to the technical field of moving target detection in the field of pattern recognition, and discloses a moving target detection method in a slowly changing moving background, in acamera video, each frame of image can be segmented into a foreground target and a background, and supposing that the resolution of the camera video is p * q, the moving background sequence is B1, B2,..., Bk, and the image sequence is X1, X2,..., Xm. The method for detecting the moving target in the slowly-varying moving background comprises the following steps of: firstly, performing data dimension reduction by using low-rank sparse decomposition, taking a video sequence as a subspace set, proposing a calculation model by using sparse approximation recursive representation in combination withdictionary sparse representation, and then calculating a moving target object by using a background difference method; therefore, the operation complexity is effectively reduced, and the operation time is shortened; secondly, according to a sparse code migration idea, the operations are carried out in a down-sampling space, so that the operand and the requirement on memory storage are further reduced, and meanwhile, the method is suitable for multi-scale target objects and can overcome the influence caused by a very large abnormal region.

Description

technical field [0001] The invention relates to the technical field of moving object detection in the field of pattern recognition, in particular to a method for detecting moving objects in slowly changing moving backgrounds. Background technique [0002] In the field of computer vision and pattern recognition, moving object detection is a very important and active research direction. At present, moving object detection has been widely used in many fields, such as industrial monitoring, traffic control, robot navigation, behavior recognition and intelligent monitoring, etc. . [0003] In order to be able to solve problems such as background changes, lighting changes, and moving shadow interference, and to accurately, real-time, and efficiently detect the target object in the camera, relevant researchers have carried out a large number of algorithm model research and improvement. In the early days, due to the development of compressed sensing and sparse representation theory...

AI Technical Summary

Problems solved by technology

[0005] (1) The method based on background differential modeling has good performance for smooth camera motion and specific PTZ situations, but due to its moderate complexity, it is difficult to be widely used in real-time applications

[0006] (2) The method based on low-rank and sparse decomposition can use the subspace learning model to reduce the dimensionality of video data and obtain a low-dimensional background model, so as to efficiently detect moving objects and be used for the detection of dynamic objects in moving cameras. It has outstanding advantages, but its computational complexity is relatively high, and its real-time performance is relatively poor.

Images