A hue-saturation histogram difference method to vehicle detection

To prevent moving shadows being false detected as moving vehicles, this paper presents a Hue-Saturation Histogram Difference (HSHD) method for vehicles detection. In the method, H-S histogram of each frame in the detection area is counted firstly. Then, comparing the differences between histograms of two adjacent frames, the peak of differences may stand for a car. Experiments on videos shot in daytime and nighttime show HSHD algorithm is feasible and robust, and the detection accuracy rates are over 97% in daytime, and over 90% even in nighttime.