The Recognition and Enhancement of Traffic Sign for the Computer-Generated Image

Image recognition in image understanding is very challenge research topic. However, the study of recognition and enhancement of the specific target for the computer generated image is very less. Compared with the natural image, this cause may be the characteristics (simple in overall) and the scene of applications (the 3D roaming, the game and so on) for the computer-generated image. For recognition and enhancement of the specific target (such as traffic sign) in the computer-generated image, the difficulty is how to accurately recognize and enhance the specific target, and maintaining other information of object do not been changed. In this paper, adopting stepwise refinement method attempts to solve the problem. The proposed method is based on three steps including image preprocessing, image recognition and enhancement. The experimental results show the method can be very good to deal with such issues. The research result shows that the image recognition and image enhancement is still not simple process. Therefore, we imagine that image recognition for complex real scene image also isn´t simple issues. But the research idea of stepwise refinement which is proposed in the paper provides a methodological reference for complex image recognition.