Research On X-ray Image Processing Method In The Security System

With the improvement of people’s living standard and the electronic commerce recently, the post has got great development in logistics industry. As a result of security check of the post closely relates to people’s personal and property security, so it has received the widespread attention. However, due to the related technology is not perfect, leading to the rates of both miss and false detection are high in security check.Therefore, in order to prevent some accidents, there is important research value and practical significance to improve the efficiency and accuracy of detection of dangerous products.Our research work concentrates on the X-ray image which is generated by the X-ray security inspection machine. Through combining the research on the processing method of the X-ray image with the recognition principle of dangerous products,we want to improve the detection accuracy of security inspection machine.The X-ray security inspection machine structure and working principle and X-ray image formation principle is researched and discussed in detail in this paper.According to the color characteristics of the X-ray image, the Ostu method and HSI color space were used for processing the X- ray images, and segmented suspect dangerous area from the original image.The theory of generalized Hough transform was researched, and used for image matching. In order to increase the matching rate of rotation image, we has promoted the generalized Hough transform in the paper.We proposes a improved method combining the mathematical statistics theory and generalized Hough transform in this paper, and introduces a distance threshold l which adds a matching standard factor, effectively improves the matching precision of dangerous products.In the MATLAB Tools, in order to detect dangerous products, the paper selected many X-ray images to do the experiment. The result showed the improved method put forward in this paper was better than traditional generalized Hough transform for the matching rotation image.The improved method can increase the accuracy of dangerous products detection. It proved the method proposed in this paper is effective.