Reserch Of Keytechnologies In Supervision System For Motor Safe Inspect Agencies Based On Image Retrieval

With the social development and improvement of living standards, The demand for motor has also further strengthened. Therefore, the motor safe inspect agencies also increase their workload, but now there are still many imperfections in the existing motor safe inspect agencies. For their own economic intersets many motor safe inspect agencies ignore the regulations of the country. They use various opportunistic approach to improve the qualified rate of motors in order to attract the owners to come back and they will gain hifher profits. So we need to establish supervision system based on computer image technology for motor safe inspect agencies, in order to strengthen supervision of motor safe inspect agencies.Based on this background, this paper analyzes the necesstiy of establishing the supervision system for motor safe inspect agencies and key technologies. We design the overall structure of the supervision system for motor motor safe inspect agencies. Based on experimental process, we will establish the supervision system for motor safe inspect agencies based on the image retrieval technology and solve the problem of motor safe inspect agencies.In the part of the video sampling and image processing, this paper completes video surveillance for motor safe inspect processing by the surveillance camera, video capture cards and others. Considering to the actual situations, the causes of image disfiguration formation in the process of the image sampling was analyzed, and we propose methods for eliminating these defects, such as gray stretch, median filter and gradient sharpening. We use the way of horizon and vertical projection to complete the location of license plate and also use the way of the Hough Transform to correct the inclined license plate, according to characteristics of the license party.In the part of the segmentating characters on the license plate, this paper consider to use a traditional character segmentation method based on vertical projection and we also propose the way of dealing with the adhesion and fracture characters, considering to the actual situations of the license plate. In the part of the character recognition, the features of gray, grain and profile are applied as input parameter to train the neural network, and recognition result is analyzed.In the part of the image retrieval, according to the requirements of the supervision system, this paper achieves that it can search key frame extraction of images from video process and it also completes to retrieve images from the video process.In this paper, through learning about the knowledge of the digital image processing, neural networks and image retrieval algorithm, we establish the supervision system for motor safe inspect agencies based on the image retrieval technology. At last, we verify the feasibility and universality of the supervision system by the image recognition and retrieval experiments.