Foreground detection method based on adaptive background update and selective background update

Abstract

The invention discloses a foreground detection method based on adaptive background update and selective background update and belongs to the field of image processing technology. According to the method, first, image frames of a to-be-detected video stream are converted into grayscale images and HLS images; if the current image frame is the first frame of the video stream, the grayscale image of the current image frame is used as a background model; for the image frames except the first frame, if the number of the image frames does not exceed a threshold, the background model is updated continuously based on the grayscale image of the current frame till the number of the image frames exceeds the threshold; and when the number of the image frames exceeds the threshold, the foreground imagedetection processing step is entered, wherein a first foreground image is obtained based on the background model and the grayscale image of the current frame, then a frame difference method is adoptedto acquire a second foreground image, a union set of the first foreground image and the second foreground image is taken to obtain a third foreground image, the third foreground image is processed through a color-normalized cross-correlation coefficient, and therefore a pixel-level foreground target is obtained. The method has a good detection effect on a moving target and a static target and hashigh robustness.

Description

technical field [0001] The invention belongs to the technical field of image processing, and mainly relates to the field of foreground object detection. Background technique [0002] Foreground detection, as an important part of intelligent video surveillance, has received extensive attention in recent years. The performance of foreground detection is critical for subsequent processing such as object classification, object tracking, and behavior understanding. It has become one of the key research directions of research institutions. In recent years, many effective methods have been proposed. Although the detection speed is fast and the background modeling is simple, the target residual phenomenon is serious, especially for slow moving and stationary targets. The detection effect is not satisfactory. [0003] For example, Barnich O, Van Droogenbroeck M proposed a ViBe method called visual background extraction in "ViBe-a powerful technique for background detection and subt...

AI Technical Summary

Problems solved by technology

However, since the algorithm itself considers the ablation of the Ghost area in the background, it is completely unable to effectively detect the slow moving speed and stationary targets.

Images