Target ghosting detection and edge propagation suppression algorithm

Abstract

The invention provides an algorithm which is based on a ViBe algorithm, introduces an iterative self-organizing analysis algorithm and an edge propagation suppression algorithm to realize suppression of a detected target ghosting and prolongs a slow moving target to be fused into a background sample model. The implementation of the algorithm ghosting suppression comprises the steps of calculating an optimal segmentation threshold value of a current frame through an introduced iterative self-organizing analysis algorithm (ISODATA), discriminating the current frame in combination with a ViBe algorithm, and performing secondary discrimination on foreground pixel points to suppress the ghosting. And for a target slowly fused into the background sample model, the invention includes calculating a high threshold value of a Canny operator of a background region by using an Otsu algorithm to judge whether the background region has edge features, and if the edge features do not exist, implementing background updating according to a background updating strategy of a ViBe algorithm; and finally, completing target detection through morphological processing.

Description

technical field [0001] The invention belongs to the technical field of image processing, relates to target recognition in video signals, and is an algorithm for suppressing ghost images of targets in video signals and suppressing target edge propagation based on a ViBe algorithm. Background technique [0002] Moving target detection is a key preprocessing technology in computer vision, which provides the basis for subsequent target tracking and recognition, action behavior analysis and other applications. Traditional target detection algorithms include: frame difference method, optical flow method and background subtraction method. The background subtraction method is currently the most commonly used target detection method. Its basic principle is to establish a background model and compare the current frame with its background model to determine the foreground target. [0003] In 2009, Olivier Barnich proposed the Visual Background extractor modeling method in "ViBE: A pow...

AI Technical Summary

Problems solved by technology

However, the ViBe algorithm also has some problems. First, the algorithm uses the first frame of the video sequence to complete the background modeling. Although the response speed is fast and it can meet the real-time requirements, when the first frame of the image contains moving objects, the algorithm will fail. The background of the corresponding area is misjudged as the foreground, forming a ghost phenomenon that affects the accuracy of target detection

Secondly, because of the spatial random propagation mechanism of the algorithm, slowly moving objects will be integrated into the background library, resulting in erosion, resulting in incomplete detection targets

[0004] In response to the ghosting phenomenon, in 2017, Wu Jiansheng et al. published "Vibe Moving Object Detection with Dynamic Threshold" in "Computer Engineering and Application" and proposed an improved algorithm for dynamic threshold, using the Otsu algorithm to calculate the segmentation of each frame of image Threshold, and then distinguish the ghost pixels, but when the variance between the foreground and the background is small, the image processing effect is not good

In 2014, Li Xiaojuan et al. proposed to detect ghosts by comparing the histogram distribution and the pixel change rate to judge the target similarity of the foreground target in the "Ghosting Suppression Method of Target Similarity Measurement" published in "Computer Application Research", but its The computing efficiency is low, and the real-time requirements of the monitoring system cannot be guaranteed

Since the high and low thresholds of the traditional Canny operator are determined artificially, they do not have general applicability

Images