Vehicle Color Recognition Based On Modified Color Histogram And Car Window Detection

Vehicle information recognition is a key part of the intelligent transportation system(ITS),and the color is one of the main features of the vehicle identification.The most common method of vehicle color recognition is color histogram method,the traditional color histogram method has high color feature dimension,and the uniform quantization method does not take into account the defect of human eyes,so it has a poor performance on vehicle color recognition.In this paper,we analyzed the traditional histogram description method and color space model,and propose a modified color quantization method and color histogram method for vehicle color recognition.Car window detection is applied to select the region of interest indirectly,which improved the accuracy of modified color histogram method.The main contents and innovations of this paper are as follows:1.A modified color histogram method is proposed to achieve the vehicle volor recognition in this paper.The algorithm solves the shortcomings of the traditional histogram.By analyzing the traditional histogram description method and color space model,the color of the car is quantified into 10 categories in Hue?Saturation?Value(HSV)color space,then a modified color histogram is extracted and applied in vehicle color recognition.The results show that the method can effectively identify the color of vehicle in the picture.2.In order to suppress the influence of the window and background,car window detection is applied to select the region of interest indirectly,which improved the accuracy of color histogram method.3.Experiment and verification.Simulation and comparison are conducted on the pictures of different colors,verified the effectiveness of this modified color histogram in the application of vehicle color recognition,which can improve the accuracy of vehicle volor recognition.