Video motion shadow detecting method based on lighting compensation

Abstract

The invention provides a video motion shadow detecting method based on lighting compensation. The method includes: using a background differencing method to obtain the foreground area with motion shadows, using color features to extract candidate shadow areas, randomly sampling the areas to estimate the parameters of lighting compensation, and differencing the compensated background area with the foreground area to obtain the motion shadow detecting result. The method has the advantages that any scene assumption can be avoided, and manual intervention is not needed; compared with other multi-feature shadow detecting methods, the method can fuse multiple features reasonably, instantaneity and accuracy are achieved, and the method is well applicable to scene and environmental variations.

Description

technical field [0001] The invention belongs to the field of video image processing, and relates to a motion shadow detection method capable of adapting to scene changes, in particular to a video motion shadow detection method based on illumination compensation. Background technique [0002] Moving target detection (foreground detection) is a basic and important task based on computer vision monitoring. Accurate and effective detection results are the prerequisite and foreshadowing for subsequent advanced applications such as target tracking, target recognition, and scene understanding. However, due to the influence of the environment and the actual scene of video surveillance, the result of moving object detection often contains a lot of false information, and moving shadow is the most common form. Motion shadow is a kind of projection caused by the light occlusion of some areas in the scene due to the object's occlusion of light during motion. The significant difference b...

AI Technical Summary

Problems solved by technology

Local texture and regional texture methods have relatively poor detection results for regions with inconspicuous texture features, and need to continuously adjust the size of the region according to scene changes to obtain accurate results

In addition, there are some methods that combine multiple features such as color and texture, but there are still many problems in how to carry out effective information fusion and real-time performance.

The above-mentioned algorithms mostly rely on the scene and human interaction, and the features used are relatively single, the adaptability to scene changes is poor, and the accuracy and real-time performance need to be further improved

Images