Research Of Image Enhancement Technology Based On DSP

In modern scientific research, military technology, industrial and agriculturalproduction, medicine, meteorology and other fields, the image information to be usedby more and more people to understand things and solve practical problems. In real life,the images obtained by the imaging system will have some degree of degradation,resulting in the loss of image information. Shooting under low visibility conditions,the image contrast and color in the scene under the influence of the fog be changed ordegraded, the image contains many of the features are covered or blurred, so the scenecan greatly reduce the degree of recognition, which requires image enhancement andrecovery technology to improve people’s visual effects, image processor with the rapiddevelopment of video image processing system has been widely applied, due to thecharacteristics of digital image itself, the computer or real-time hardware processinghave high precision, even can handle complex non-linear processing, and have ability offlexible alternative. To study the image enhancement algorithm and implement it on theembedded platform have a good practical value.This paper mainly study on the use of image processing techniques to fog ordegraded image for image restoration and enhancement, research contents include thefollowing:(1) The classical image enhancement algorithms were studied, and simulate thisalgorithm respectively.(2) Describe the algorithms of Single image haze removal using dark channel priorand take a detailed explanation and analysis for Matting Technology.(3) Introduce an algorithm of Fast visibility restoration from a single color or graylevel image algorithm, the algorithm complexity is only a linear function of the totalnumber of pixels, and can both handle color or grayscale images.(4) Build DSP platform, Transplant Single image haze removal using darkchannel prior algorithms to DSP platform for real-time video processing, because of thehigh complexity of this algorithm, real-time is not well, so we change to use Fastvisibility restoration from a single color or gray level image algorithm, afterprogramming optimize, finally get a better real-time and treatment effect.