Research On Technology Of Recognition Of Car Face Major Information

With progressing of technology of digital image processing and video surveillance equipment, intelligent transportation systems attracted more and more people’s attention. Recognition of car face major information as an important part of intelligent transportation systems, its related technology research had important practical significance, and had a wide range of applications in the regulation of traffic control, traffic statistics, security and public security systems and other areas. The main content of this paper is the recognition of car face major information, including license plate recognition, body color, and the color of the car license plate and so on. Firstly, the image preprocessing methods, space median filtering and morphological processing were introduced, which provided the foundation for subsequent recognition works. The recognition system of car face information was divided into six parts, such as license plate positioning, license plate color recognition, skew correction, character segmentation, character recognition, body color recognition. The main innovation of this paper and can be grouped into the following four points:1. A license plate location method based on a special gradient operatorDrawing on the license plate position method based on gray grayscale morphological image processing techniques, this paper proposed a license plate location method based on gray gradient image. License plate image features include color and texture features, the license plate characters and background have the greatest contrast in the yellow channel or blue channel. The pixel of license plate has obvious contrast at the boundary between background and character. Therefore, this paper established a gray image of yellow and blue color channels. A special Operator was proposed to calculate the gradient image. The Operator was designed as the neighborhood poor value in the level direction to enhance plate features. Then, the gradient image was processed through the gray morphology method to extract features of license plate and locate the license plate image. Through comparative analysis of the experiment, this enhances the feature extraction effect of image at night. The algorithm is simple and adaptable.2. License plate image correction and character segmentationFor tilt plate image correction, the classical algorithm based on Hough transform has shortcomings of poor applicability in vague and polluted image. A linear regression model was used to estimate the tilt angle of the license plate in this paper. The sample data is the plate boundary coordinates of characters’connected domain within the license plate region. For character segmentation,9cut lines in the vertical direction were established to cut the7characters according to the standard of license plate character size and symmetry. We made the sum of cutting lines’ pixels as the objective function proposed an optimization model for character segmentation. The effectiveness of the method was verified through simulation experiments and theoretical analysis. The method in this paper is an effective solution to cut license plate sticking characters image.3. A license plate color recognition method based on K-means AlgorithmFor the license plate color recognition, this paper proposed a color recognition model based on K-means clustering method. For the license plate image, a shrinking method was used to select the sample of image pixels for color recognition. According to the spatial characteristics of the license plate pixels, we calculated the color pixel values of background class and character class by K-means clustering. A color correction model was proposed to correct the image color according to relationship between brightness and color channels. And the color values were adjusted by this way, which improved the accuracy of recognition result. The binarization results were enhanced with the binarization method based on clustering results. The validity and applicability of the algorithm was verified through a large number of license plates image color recognition.4. A method of color correction and body color recognitionFor the body color recognition, there is a problem that the image color was changed by the light of car on the night and the image brightness was not uniform. A pixel model was proposed to estimate the information of color change, which can be used to correct image color and contrast. Combining the RGB space and HSV space, a method of color recognition was proposed to recognize the car body color after image color correction. By the experiments of night image processing and the comparative analysis of traditional image color correction algorithm, the effectiveness of the method in this paper was identified.