New Approach in Image Compression: 3D Spiral JPEG

This paper presents a new JPEG-based lossy image compression method based on three-dimensional formation of the original image by spiral order scanning and 3D discrete cosine transformation (DCT). The proposed spiral scanning results in similar information blocks to be in the same 8x8x8 cube. The proposed quantization table and method for converting a 3D image cube into a ID array provide better coding efficiency in run length coding. Hence, a better performance in a wide range of images is obtained. The performance of the proposed compression method is measured over various images, and it is observed that the proposed method has better performance than that of conventional JPEG compression especially in low and in high bit rates. In addition, the proposed compression scheme has generally superior performance in images where there is substantial amount of background.