Study On Key Technologies Of Color Image Processing

With the development of information technology, color images are applied more widely. Compared with grey image, more visual information is brought by color image. Color image processing has become an important study domain. Because of the complication of color representation in color image, the grey image technologies can't be directly applied in color image processing. How to utilize color information sufficiently is most important in color image processing.Color optics and human color vision characteristics are studied and summarized at first in the dissertation. Based on that, various main color spaces, which have been presented by now, are classified and analyzed. And the transform relationships among those spaces are discussed.A novel definition about color distance measurement is given. By analyzing the value relationships among parameters in RGB and HSI color space, color similarity coefficient is defined to measure degree of color similarity between two kinds of color, then an algorithm of measurement of color similarity is brought forward. It is simple and realized more easily.Based on BP neural network, an algorithm of color classification for color image is presented. There is a nonlinear relationship between the parameter values in RGB color space and color classification according to human color vision characteristics. BP neural network has efficient learning capability, simulation ability for cognizing system of human, and very good capacity of nonlinear classification. Based on BP neural network as a classifier to classify color characteristic vector extracted, a series of measure, such as characteristic vector standardization, fuzzy pretreatment and most membership output are applied to classify color vector in RGB color space.An algorithm of selective image filtering based on impulse noise detection is presented. Impulse noise is a frequent noise in image. It affects the image quality badly. Removal of impulse noise is very important for image processing. According to the image characteristics of impulse noise, two methods respectively based on scalar detection in multiple color channels and vector detection are proposed. Based on the...