| This thesis provided two new methods for image quality assessment and artifact removal of JPEG images. In chapter 2, we proposed a no-reference quality assessment method for JPEG images. The chapter presented a no-reference quality assessment algorithm for JPEG compressed images (NJQA). Our method did not specifically aim to measure blockiness. Instead, quality was estimated by first counting the number of zero-valued DCT coefficients within each block, and then using a map, which we called the quality relevance map, to weight these counts. The quality relevance map for an image is a map that indicates which blocks are naturally uniform (or near-uniform) vs. which blocks have been made uniform (or near-uniform) via JPEG compression. Testing on various image-quality databases demonstrates that NJQA was either competitive with or outperforms modern competing methods on JPEG images. In chapter 3, we proposed an algorithm for JPEG artifact reduction via Guided filtering and local edge regeneration. This chapter presented a new approach that deals with artifacts in JPEG compressed images. Our algorithm addressed all three types of artifacts which are prevalent in JPEG images: blocking, and for edges blurring and aliasing. We enhanced the quality of the image via two stages. First, we removed blocking artifacts via boundary smoothing and Guided filtering. And second, we reduced blurring and aliasing around edges via a local edge-regeneration stage. We compared the proposed algorithm with other modern JPEG artifact-removal algorithms. The results demonstrated that the proposed approach is competitive, and can in many cases outperforms, the competing algorithms. |
