The Research On Image Enhancement Theory And Application Based On Human Visual Property

Human Visual System is the most efficient, accurate and fastest image processing system in the world currently, because human eyes have some particular visual properties. Some properties such as Color Constancy and Lightness Constancy are closely related with image enhancement. Research on Human Visual Property is significant for image processing, especially in image enhancement fields. Domestic and foreign scholars obtain better image enhancement effects mainly by simulating Human Visual Property. Because of color and lightness constancy, HVS can identify and process targets fast, correctly and steadily. Therefore, many current excellent algorithms are based on these properties and are still evolving.In this thesis, Human Visual Property is the major research point. Based on the analysis of Color Constancy, Lightness Constancy and Retinex theory, new enhancement framework and calculation model are proposed and applied in foggy image enhancement and image enhancement in compressed domain to verify their effectiveness. The main content of this thesis is shown as follow:1. This thesis researches Human Visual Property theory deeply, including two properties: Color Constancy and Lightness Constancy which are closely related with image enhancement. The limitation of Retinex and other enhancement algorithms are pointed out based on the analysis of the research and theory. This provides the theoretical basis and ideas for research on enhancement framework and calculation model.2. A new enhancement model which can preserve Color Constancy and Lightness Constancy at the same time is proposed and applied in foggy image enhancement. The degradation of foggy images is non-linear with the depth of field, a new adaptive Retinex algorithm in foggy image enhancement is proposed at first. The new algorithm adjusts the parameters automatically according to the strong degree of fog reflected by the human visual area which image pixels belong to. Based on this a new calculation model of image enhancement is built. The model combines the results enhanced by adaptive Retinex algorithm and CLAHE algorithm with different weights adaptively, and obtains farther enhanced images with both Color Constancy and Lightness Constancy to make foggy images clear.3. A new image enhancement algorithm in the compressed domain is proposed. The new algorithm enhances the compressed images by dividing DCT coefficients into the incident component (DC coefficients) and the reflection component (AC coefficients) based on Retinex theory. The DC coefficients are adjusted to compress the image dynamic range, and the AC coefficients are adjusted to enhance the details. The block artifacts will be suppressed with a threshold method. Processing both the incident component and reflection component will make full use of low and high frequency information of original images, which can achieve the purpose of details enhancement, color preserving and artifacts suppressing.