Background removing method for restricting influence of sudden changes of light in video analysis

Abstract

The invention discloses a background removing method for restricting the influence of sudden changes of light in video analysis and relates to background removing methods. The method comprises the steps that a video streaming input frame is processed by a background removing module, so that all foreground points and the information of upgrading history background Ib of the frame are obtained; the average gradient of an MxN neighbourhood of each foreground point (x, y) in the frame is calculated and is compared with a threshold Tg; if the average gradient is smaller than Tg, subsequent steps do not need conducting, and the point is judged to be a background point directly; gradient-based texture similarity insensitive to light changes is selected as a judging basis for judging whether a certain pixel is foreground; when the similarity S (x, y)) Ts in the pixel point (x, y) is obtained through calculation, the pixel point (x, y) is judged to be a background point. According to the background removing method, subsequent process is added on the basis of existing background removal, so that virtual alarm caused by quick changes of light is removed.

Description

[0001] technical field [0002] The invention relates to a background removal method, in particular to a background removal method for suppressing the influence of sudden light changes in video analysis. Background technique [0003] Background removal is one of the most important steps in intelligent video analysis. Most of the pixels in the video are unchanged for a long time. Background removal can remove the unchanged background in the video and leave the changed foreground. Mixed Gaussian model background removal is currently the most popular method for background removal. The mixed Gaussian model uses K (basically 3 to 5) Gaussian models to characterize the characteristics of each pixel in the image, and updates the mixed Gaussian model after a new frame of image is obtained, using each pixel in the current image and the mixed Gaussian Model matching, if successful, it is determined that the point is a background point, otherwise it is a foreground point. Looking at...

AI Technical Summary

Problems solved by technology

[0006] However, one of the main disadvantages of the hybrid Gaussian model background removal method is that when the light changes rapidly, such as the sun coming out of the clouds in cloudy weather, the background learning cannot keep up with the rapid light changes, resulting in a large area of ​​false alarms.

[0007] At present, all self-learning background background removal algorithms can deal with slowly changing scenes or periodic motion, and have a certain suppression of noise, but lack the ability to deal with fast light changes, shadows, and dynamic textures.

Images