Image Enhancement Algorithm And The Application Based On Mathematical Morphology

Image enhancement as the preprocessing stage of image processing is an important branch of it. In general, influenced by acquisition equipment, shooting environment, transmission method and other factors, image appears fuzzy, contrast lower, and the main purpose of image enhancement is to improve image quality to meet needs of practical applications. With the extensive using of color images,requirement of color image enhancement technology is becoming higher, different from grayscale image enhancement, enhancing color image, not only to stretch image contrast, enhance image details, but also to maintain brightness, avoid color distortion and other problems.The application of mathematical morphology in image processing becomes a research hotspot in recent years. Theory of mathematical morphology algorithm is simple and its operation speed is fast, morphology top-hat transform can extract bright features and dark features of image. In this thesis, combing with the advantages of morphology algorithm, an anti-noise enhancement method of noised grayscale image and color image enhancement method is studied, the main contents are as follows:Firstly, the thesis briefly has introduced the significance of image denoising and image enhancement, several traditional denoising and enhancement algorithm are described, then the simulation experiments of these algorithms are carried out.Secondly, the theoretical knowledge and basic algorithm of mathematical morphology is introduced detailed and systematically, the results of simulation experiment are analyzed. Based on the effects of dilate and erode operation, a multiform dilation erosion filter is designed, for less transform times, which can effectively restrain noise, keep details and edge information of image. In addition,make improvement for the existing multiscale top-hat transform, so that the enhancement effect is more ideal. Combining these two steps, a multiform multiscale mathematical morphology denoising enhancement method for grayscale image is proposed. The simulation results show that,when noise density is smaller,the proposed method can completely suppress noise, and the enhanced image have clear details, mass information and good visual effect.Then, the multiscale top-hat enhancement algorithm in grayscale domain is extended to the color space, for operation complexity, information loss and other problems caused by spatial transform in color image processing. In this thesis, a method based on mathematical morphology top-hat transform, and using standard deviation weight ratio of each component as regulatory factor is proposed to enhance color image, in RGB space. Experiment result indicates that the method can effectively enhance color image contrast, highlight details, avoid excessive enhancement, and can keep color and brightness of the original image well.Finally, the method is applied to anterior segment images, and compared with other algorithms, results show that, method proposed in this thesis has fast operation speed and the enhanced grayscale anterior segment images have clear details, obvious pupil information. The enhanced color anterior segment images can maintain color,enhance detail features, and have a good visual effect. Which is helpful for computer or medical personnel quickly make an accurate diagnosis plan.