Foreground detection method based on background model and frame difference

Abstract

The invention discloses a foreground detection method based on a background model and frame difference, relates to image processing and particularly relates to the foreground detection field. For a problem that a background is liable to be taken as a foreground to form a large shadow or a large spot existing in a foreground detection method based on background modeling during light change in the prior art, a frame difference method is utilized for insensitivity of light to propose the foreground detection method based on the background model and frame difference. According to the foreground detection method based on the background model and frame difference, statistics of background pixels is carried out through utilizing background modeling, a real-time background pixel model is formed in combination with inter-frame difference, the method mainly comprises steps of background initialization and background model update. An inter-frame difference multi-neighborhood pixel discrimination method is merged into update of the background model, and a light insensitivity problem existing in the background modeling method in the prior art is solved.

Description

technical field [0001] The invention relates to image processing, in particular to the field of foreground detection. Background technique [0002] The main purpose of foreground detection is to extract the changing areas or moving objects in the image sequence from the background image. In intelligent video processing, for the processing of surveillance video, the foreground image must be detected from the background in order to further analyze the moving target. Therefore, foreground detection is a key technology in video processing. It has important applications in identification and tracking. Because the background of the video image will change with the influence of light, camera shake and other factors, it is difficult to accurately separate the background from the foreground. At the same time, for the processing of massive videos, the speed of foreground detection directly affects the video processing speed, so it is urgently needed A fast and accurate video foregro...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the defect that the existing foreground detection method based on background modeling is easy to detect the background as the foreground to form large shadows or large "spots" when the light changes, the present invention uses the insensitivity of the frame difference method to light, and proposes a method based on Background model and inter-frame difference foreground detection method. The basic idea of ​​this method is to use background modeling to count background pixels and combine with inter-frame difference discrimination to form a real-time background pixel model, which mainly includes background initialization and background model update.

Integrate the multi-neighborhood pixel discrimination method of inter-frame difference into the update of the background model to solve the sensitivity problem of the existing background modeling method to light

Images