Perceptual based image quality assessment and enhancement

The Human Visual System (HVS) behavior is investigated and its response to different amount of blur in an image is studied. A no-reference objective sharpness/blurriness metric is suggested for quality assessment based on a new concept referred to as the Just Noticeable Blur (JNB). An HVS model is derived predicting the perceived quality of the blurred images having a key advantage of predicting sharpness/blurriness over a wide range of images with different content. An evaluation of existing sharpness/blurriness metrics is performed and a comparison is drawn showing the superiority of the proposed metric compared to the existing ones specifically when applied to natural images with different characteristics such as texture, edges and smooth regions. Subjective tests are also performed and statistical methods are used to assess the correctness of the proposed metric. It is shown that the metric correlates strongly with the Mean Opinion Score (MOS).;In addition, the HVS masking characteristics are exploited for enhancing images by using a set of low resolution images to reconstruct a high resolution one using super-resolution. A selective scheme is presented to reduce the complexity of popular super-resolution algorithms while maintaining the desired quality of the enhanced images/video. A perceptual Human Visual System (HVS) model is proposed to compute the contrast sensitivity threshold for a given background intensity. The obtained thresholds are used to select which pixels are super-resolved based on the signal activity. The main idea of the proposed algorithm is to predict whether local edges of the images are visually perceived. This is accomplished by estimating the contrast sensitivity threshold locally over a block. Next, the absolute difference between each pixel and its neighbors is computed and compared to the threshold upon which a decision is made to include the pixel in the SR estimator for the next iteration or not. The perceptual model is integrated into a MAP-based SR algorithm as well as a fast ML estimator based on a two-stage algorithm. Simulation results show up to 50% reduction on average in computational complexity with comparable SNR gains and visual quality.