Research Of Railway Freight Detection Based On Image Processing

Now railway freightage is the main materials transportation in our country. With the rapid development of high-speed railway, the safety management in railway freightage is much more important than ever before. The research in image processing and pattern recognition used for the railway freight detection system is of great significance and originality.In this paper, the image enhancement algorithms based on Retinex theory were deeply studied. Different from the traditional image enhancement algorithms, the Retinex algorithm can achieve good results in edge enhancement, color constancy ,dynamic range compression of the images and so on. The details of the image in the shadow area can be emphasized and the contrast of the image can be enhanced. This paper mainly introduces and implements the Single Scale Retinex (SSR) and Multi-scale Retinex (MSR) image enhancement algorithms. Base on the existing MSR algorithm, an improved MSR algorithm was proposed. Also, combined with Mean shift smooth filtering, the improved MSR gray image enhancement was formed based on Mean shift smooth filtering. Finally, two aspects about the effects of the images were considered both in human observer and objective parameters. The results showed that the proposed algorithms in this paper can not only increase the dynamic range compression of images and improve the image brightness, but also maintain the image details and improve the clarity of the image. The information entropy of the image was greatly increased through using the algorithms, and thus much more image information was obtained.In the location and segment algorithms of the railway freight, the proposed algorithms have been used to enhance the images obtained from the environment with poor lighting conditions. The maximum information entropy method was used to determine the segment threshold of the image. The object position can be obtained by using the minimum rectangular bounding box. The simulation results showed that the proposed algorithms in the paper enhanced the images of railway freight distinctively and improve the positioning accuracy.