Motion detecting device for estimating self-adapting inner core density and method therefor

Abstract

The invention provides a motion detecting device for estimating self-adapting inner core density and a corresponding method. The device of the invention comprises a receiving unit, an initiating unit, a probability calculating and statistical unit and a classifying unit, wherein, the receiving unit is used for receiving an input video frame; the initiating unit is used for initializing a background model and an inter-frame differential background model; the probability calculating and statistical unit is used for calculating estimated probability of the core density of each pixel of the present video frame according to a background sample; the probability calculating and statistical unit is also used for forming an estimated probability histogram of the core density of the pixel of the present frame by statistic; the classifying unit is used for solving classification threshold by utilizing the estimated probability histogram of the core density in a self-adaptation way; if the estimated probability of the core density of a certain pixel is smaller than the threshold value, the pixel is determined as a foreground pixel; otherwise the pixel is determined as a background pixel.

Description

technical field [0001] The invention relates to background model generation in the field of video analysis, and adopts an adaptive threshold value selection method for pixel classification. According to the technical solution of the present invention, the selection of the threshold can be performed adaptively, and can adapt to different scenarios. On this basis, the present invention further proposes a background update model to update the background. At the same time, the present invention also utilizes the frame difference background model and kernel density estimation classification results to detect sudden changes in the background, which better solves the problem of misdetection of moving objects when the background changes suddenly. Background technique [0002] Separating moving targets from video sequences is an important research content in computer vision, which can be applied to traffic monitoring, human behavior recognition and human-computer interaction and oth...

AI Technical Summary

Problems solved by technology

However, in reference [2], the threshold is selected by a given false positive rate (False Positive), which requires prior knowledge of the scene, and manual intervention is required for different scenes, reselecting

Images