Moving vehicle detection in digital image sequences is one of the key technologies of Intelligent Transportation Systems (ITS). However, problems arise due to the shadows of sunshine in daytime and the illuminations of vehicle headlights in nighttime. To begin with, a new autoregression algorithm based on Gaussian Distribution hypotheses is proposed for background estimation. Furthermore, a pivot approach to eliminate shadows and illuminations from the foreground, which is the difference between dynamic image and background image, is investigated and studied. And in this proposed approach, image textures are extracted by fast wavelet transform (FWT) which is designed for discrete signal while grey level co-occurrence matrix (GLCM) is employed to measure and analyze the extracted textures. Subsequently, shadows and illuminations can be segmented since their textures differ from those of vehicles. Experiment results in real traffic scenes reveal that the techniques presented in this work are effective and efficient for vehicle detection.Index Terms -Vehicle detection, Texture analysis, Elimination of shadow and illumination, Fast wavelet transform, Gray cooccurrence matrix.
In outdoor vehicles detection system based on video signal processing, the shadow of the vehicle detection and removal is a key link. In this paper, a novel vehicle's shadow detection and removal algorithm is proposed. Firstly, the texture autocorrelation is used to pre-extracted the shadow of the vehicles. Secondly, the statistical discrimination method is used to evaluate the shadow pre-extraction results. Then the integer wavelet transform is used to re-extracted the misjudgment of the shadow area. Finally, the two shadow extraction results are combined to implement the shadow detection and removal of the vehicle. Experimental results are showed that: the method not only can accurately detect the shadow of the vehicle which is a large difference in grayscale to compare with the background, but also can better detect the shadow of the vehicle which is similar to the background in grayscale. Therefore the method solves the common false detection question of the shadow when use the single method to detect the shadow and remove it, and obtain a perfect shadow detection and removal results.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.