Adaptive Rendering Method for Low Resolution Auto-stereoscopic 3-D Medical Images

Abstract

Three-dimensional (3-D) displays have been receiving attention since they are more realistic and intuitive than two dimensional (2-D) displays. In this study, a rendering method to extract important features and enhance them for creating Integral Videography (IV) for medical use was studied and implemented. The resolution of IV mainly depends on the number of lenses which constitute the lens array. Consequently, the resolution observed from one viewpoint is not high enough for practical use and it sometimes prevents observers from understanding quickly. This is because the low resolution cannot depict high frequency image and it even becomes noise. We introduced non-photorealistic rendering (NPR) approach to the IV image. The NPR method will enable observers to understand the structure more intuitively and precisely because the method enhances important features. The idea to apply NPR methods to optimizing the rendering method of IV is conceptually unique. In this study, we proposed the feature extracting algorithm with dynamic threshold using the image depth for IV image.