An Image Enhancement Algorithm Based on Second Generation Wavelet Integer Transform

Traditional histogram enhancement algorithm to enhance image has the defect of over-bright image, amplified noise, and missing image details. In view of these defects, this paper proposes an image enhancement algorithm based on second generation wavelet integer transform. At first, use the second generation wavelet integer transform to decompose the original image with wavelet. Then, integer low-frequency subimage ca in the wavelet domain is calculated with histogram equalization to figure out low-frequency subimage ' ca . And make an evenly spaced arrangement for integer low-frequency subimage ' ca of equalized histogram between the maximum and the minimum to figure out new low-frequency subimage ' ' ca . At last, low-frequency subimage ' ' ca is reconstructed. In comparison with classic histogram equalization algorithm and experimental results of homomorphic filtering, algorithm in this paper keeps the optimal in enhancing image, suppressing noise and preserving brightness. Introduction Image enhancement is to improve the visual effects of images and make them more clear and intuitive and make it appropriate to analysis. There are many algorithms enhancing the images. The histogram equalization is much more classic and effective. Despite the obvious advantages of fast arithmetic speed and great enhancement effectiveness, it still has the following shortcomings: (1) The histogram equalization would cause poor layering sense because of the small dynamic range of the original image’s gray scale, its poor quality and uneven histogram equalization. (2) After the histogram equalization, the noise in the original is obviously enlarged. (3) If the image gray scale is close to 0, during the histogram equalization, a brighter diluted image would be output when the very narrow range of dark pixels is mapped to the output image. This would affect the basic features of the image (for instance, the average luminance would be changed; the details would be lost), which would finally influence the visual effects to enhance the image and make the applications of histogram algorithm limited . Several classic methods to enhance image Histogram Equalization Enhancement. The basic idea of histogram equalization is to transform the unbalanced histogram of the original image to form of uniform distribution. Namely, convert the input image to have the same pixel points in each gray level (The histogram output is smooth and distribution is uniform).This makes dynamic range of the image pixels wider and contrast stretched effectively so as to achieve the effect of image enhancement. Histogram 2nd International Conference on Advances in Mechanical Engineering and Industrial Informatics (AMEII 2016) © 2016. The authors Published by Atlantis Press 242 equalization enhancement has a disadvantage that if the noise image is enhanced, a little noise will be greatly enlarged at the same time. The following is a digital image histogram equalization algorithm formula. n n r p k k r / ) (  ( 0≤ k r ≤1 k=0,1,2,...,L-1 ) (1) In formula (1), L is the sum of gray levels. ) ( k r r p is probability of Class K gray value. k n is the times of Class K gray value existed in the image. is sum of all pixels in the image. New image grey value is as follows: