Adaptive background reconfiguration method based on pixel sequential morphology

Abstract

The invention relates to an adaptive background reconfiguration method based on pixel sequential morphology, which applies the pixel sequential morphology on a time dimension to guide background image extracting and updating. The adaptive background reconfiguration method based on pixel sequential morphology extracts the characteristics of the pixel sequential morphology; as independent background extracting strategies, background updating times and background updating strategies are adaptive to different morphologies, the adaptive background reconfiguration method based on pixel sequential morphology not only can extract conventional backgrounds, but also can extract periodical background according to the pixel sequences of the periodical morphology. The adaptive background reconfiguration method based on pixel sequential morphology does not need to build models for the background and moving targets in a video scene, directly reconfigures the backgrounds from a group of video images comprising movement prospects, effectively avoids mixture, and improves the accuracy for background extracting, the effectiveness for background updating, and the instantaneity for background reconfiguration.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an adaptive background reconstruction method based on pixel sequence morphology. Background technique [0002] Background reconstruction is one of the supporting works in the field of video image processing, which involves two basic problems of background extraction and background update. At home and abroad, a lot of research work has been done on background extraction and background update, and many classic algorithms have been formed. The most classic background reconstruction algorithm is the filtering method based on the time axis, which constructs the background image according to certain rules from a set of observation images in the past. Taking the mean filtering method as an example, the video image sequence is summed and averaged over a period of time to construct an approximate background image frame and complete the subsequent background update. The calculati...

AI Technical Summary

Problems solved by technology

In addition, there is another type of background reconstruction algorithm that first establishes the background model, and then adjusts the model parameters to obtain the background image, such as Gaussian distribution model, mixed Gaussian distribution model, Kalman filter method, etc. This type of method has good adaptability to illumination changes, but The calculation is complex, the real-time performance is not good, and the Gaussian distribution model cannot effectively update the background when the moving object is stationary for a long time

Generally speaking, the current mainstream background reconstruction algorithms have great differences in the accuracy of background extraction, the effectiveness of background update, the real-time performance of background reconstruction, and the adaptability of background scenes.

Images