Research On Adaptive Local Image Enhacement Technique

Adaptive local image enhancement techniques make use of the local characteristics of the image, and then combine the image contrast enhancement with edge preservation and denoising so as to achieve local area enhancement. It can improve the image quality and overcome the deficiency of the global image enhancement techniques.This thesis systematically studies adaptive local image enhancement including contrast enhancement as well as image filtering, and realize some algorithms with large computation for real-time use.The main work and researching achievements of this thesis can be summarized as follows:1. Research on adaptive local image contrast enhancement technique. In order to enhance the images with low illumination, large background area and dim target, this paper proposes the plateau histogram equalization based on logarithm transform. This algorithm can enhance the detail in images and avoid the high noise-sensitivity of Logarithm-Lee method.2. Research on adaptive local image filtering technique. In order to overcome the limitations of the existent bilateral filters, such as incapability to sharp the details and high sensitivity to noise, this paper proposes the adaptive bilateral filter based on K Nearest Neighbors. The ability of this algorithm for filtering and detail- preservation is comparable to the non-local means algorithm while with much lower computation complexity.3. Research on adaptive local image enhancement technique based on partial differential equations. In order to overcome ill-pose of the P-M model and the filtering performance's heavy dependence on iteration times, this paper introduce a fidelity term to improve both the stability and the ability of detail-preservation.4. The adaptive local image enhancement parallel algorithms are designed and realized based on CUDA. The efficiency of adaptive local image enhancement is improved drastically.