The Research On Method Of Combining Fractional Differential And Wavelet Decomposition Used In Image Enhancement

In the modern life, image is the most direct and common way of perceivinginformation for people. In general, it is difficult for people to get rich texture andapparent contrast about weak details of image, and as people living standard is rising,the demand of image with rich detail and high resolution is urgent, then, theenhancement on image details become a way of improving image weak characteristicsand high resolution quality. This thesis will further explore and study imageenhancement method.Image enhancement, as a branch of image processing, has been proposed a lot ofmethods so far. This paper firstly studied the existing methods, starting from thedirectional characteristic of different fractional differential mask processing image andthe directional feature of wavelet decomposing image, developing a image enhancementmethod based on combining them under the principle that direction is vertical for eachother, that is FWE(Fractional Wavelet Enhancement)method. Secondly, a furtherimproved image enhancement method based on FWE method was provided, that isFOWE (Fractional Optimal Wavelet Enhancement) method. It determined a optimaldecomposition level by respectively calculating the maximum entropy of componentsgenerated by image wavelet decomposition, then, using the designed new fractionaldifferential mask to targeted process the original image and the optimal levelcomponents signal which abtained by image wavelet decomposition and reconstructionwas utilized. Finally, the results was merged so that it could furthest preserve imageoriginal feature,and also enhance texture details in those areas that gray-level did notchange evidently. Theoretical analysis shows that FOWE method can be used to furtherenlarge the effect of image enhancement. The most apparent characteristic of FOWEmethod is that it gets a optimal wavelet decomposition level by calculating a experienceformula, processing image with fractional differential method on the optimal level, infact, get better enhancement effect.This paper anlysised the fractional differentialmethod,especially reasoned the structure and transformation process of fractionaldifferential masks that would be used in image enhancement.Finally, runing above provided methods on the Matlab simulation platform, andcomparing with results of exsisting typical image enhancement methods. From thecontrast shows that the newly provided FOWE method was effective in aspect of enhancing detail characteristic of gray image.Through conmparing the results for sortsof images enhanced by FOWE method, and regularly changing optimal waveletdeconposition level in FOWE to contrast process results, reaching a conclution that it iseffective for enhancing fa ade head gray image for person or animal with thedetermination of optimal wavelet decompositon level in FOWE method. Meanwhileimplementing FOWE method in the DM642type DSP so as to lay the foundation forfurther application of portable image enhancemet device.