(A) study on contrast enhancement and quality assessment measure using contextual information

Abstract

The study on the visual quality has been extensively conducted for a long time since it can affect the performance of various computer vision algorithms. As the demand for obtaining high-quality images increases, image-enhancing techniques have attracted much attention. Among the several methods, contrast enhancement (CE) algorithm is often used to improve the image quality since contrast is an important factor in the human perception of image quality. Moreover, to develop effective and well-designed CE methods, research is needed on how to accurately predict the quality of images. In other words, the image quality assessment (IQA) can more accurately validate the superiority of CE algorithm performance and can help in the development of better CE algorithms. Therefore, this dissertation explores the study on how to improve the contrast and how to well measure the contrast quality. First, we present an adaptive CE algorithm considering both preservation of the shape of a 1D histogram and statistical information on the gray-level differences between neighboring pixels obtained by a 2D histogram. Compared with several state-of-the-art enhancement algorithms, the proposed algorithm shows highly competitive performance. Second, we propose the objective metric which can precisely predict the perceptual quality of contrast-changed images using inter-pixel contextual information. Experimental results show that the proposed metric provides a more accurate prediction of human perception of contrast change than other metrics. We further extend the CE algorithm to infrared (IR) images by using the proposed ramp distribution-based bit conversion method.