Adaptive Super-Spatial Prediction Approach For Lossless Image Compression

Existing prediction based lossless image compression schemes perform prediction of an image data using their spatial neighborhood technique which can’t predict high-frequency image structure components, such as edges, patterns, and textures very well which will limit the image compression efficiency. To exploit these structure components, adaptive super-spatial prediction approach is developed. The super-spatial prediction approach is adaptive to compress high frequency structure components from the grayscale image. The motivation behind the proposed prediction approach is taken from motion prediction in video coding, which attempts to find an optimal prediction of structure components within the previously encoded image regions. This prediction approach is efficient for image regions with significant structure components with respect to parameters as compression ratio, bit rate as compared to CALIC (Context-based adaptive lossless image coding).