Perceputal rendering algorithm

Abstract

Computer graphics has become very important as the tremendous growing speed of film, video games, visual reality and so on. However high-quality, real-time image generation usually costs a lot of computation time and physical memory. In this report, we will present a perceptual image rendering algorithm. This algorithm has implemented a suitable perceptual model which associated with Human Vision System. The perceptual model is a combination of K-mean Clustering method and Visual Importance theory. The K-mean Clustering method can generally reduced the pixel color calculation by grouping the similar color pixels. And the Visual Importance can help emphasize the parts that will be easily focused on by the human eye and omit the parts will hardly be noticed. We also use ray tracing as the basic rendering technique in order to get realistic quality. Then we use quad tree method to fulfill the ray tracing technique and at last implement with our perceptual model to form our algorithm. Our testing result also shows that our algorithm can lower the calculated pixels and save rendering time. In the meanwhile the quality of the image will not be significantly reduced.