Research On Color Enhancement Of The Low-contrast Microscopic Image

Image always is low contrast image when we use the microscope，the color isnot defined, no ease of analysis and observation.Image color enhancement techniqueis a very important issue in image processing technology. It has been widely used inindustrial production, medicine, and other fields. It is easy for researchers to have adeep study through the color enhancement techniques.In this paper, in order to realizethe achieving color enhancement effects of the microscope image, color transfertechnology and inverse convolution technology are applied to the color enhancementof the images of microscopic image. This main research contents are as follows:First, the introduction of the meaning and the purpose of this article are described,and how the color enhancement is developing in domestic and foreign is shown. Wedescribe the image of low contrast and the traditional method of improving the lowcontrast image, and offer the contrast enhancement method based on thedeconvolution algorithm. We also do some experience using the present commoncolorize method, compare and analyze the experimental results, and describeproblems in these methods.Second, colorize method based on luminance histogram partitions is offered.Curve fitting on the luminance histograms of grayscale image and the color image isperformed to analyze their feature distributions. We match these features by findingand adjusting the extreme points. Then a weighted colorization algorithm is proposedwhich overcomes the color jumping question of the color transition between regions.We also propose adaptive node selection method and fitting number selection method.Further more, we make20colorful images to gray images, then transfer it to colorfulimage.We compare it to the traditional colorization method. Our method can achievea higher value of PSNR.Third, we offer a colorful method with using the tolerance flood fillsegmentation.We use the tolerance flood fill to get the interest segmentation of themicroscope image and the color image. The interest area of the grayscale image use the pixel-by-pixel mapped rearrangement and luminance values matched method forcolor enhancement, it also needs to be distances transform in order to find the targetarea boundary transition zone, we set the number to be2, and then, we color the targetarea boundary transition zone by using the colored interest area outline. We do thisexperiment with20gray microscope images, we get the advantages anddisadvantages of the two methods of the partly showed colorize effects, and verifythat the transition zone method is able to effectively resolve the border transitionissues through the experimental verification of the method.Finally，we combine the colorize method with the deconvolution algorithm toenhance the microscopic images. we blur the true color microscope image, make it togray image, then we use the methods this article offers, and compare the effects of thismethods. The deconvolution method with the brightness value colorize method oftransitional zone achieves a higher SSIM value and a better effect through55grayscale image colorize effects from200colored images and24color image colorizeeffects from100colored images.