Research On Recognition Techniques Based On Color Histogram

Along with the rapid development of multimedia technology, computer technology, communication and Internet network, the people make more and more contact with various image informations, Color in the image information is the best characteristic for display the object characteristic, the color histogram is one of most important methods for color characteristic analysis, besides the simple computation characteristic,the method of color histogram processing also has the advantages of scale invaruance and rotate invariancerotate invariance,but the histogram is unable to obtain spatial relations information of the color composition. This article mainly has carried on the analysis for the following several aspects:1. Some key technologies are analyzed and studied based on color histogram recognition domain, including the selection of color space, the comparison of color histogram description method, the measureing criterion for the similar characteristicof color histogram.2. Regarding the extraction of color histogram characteristic, the Shafer reflection model is adopted; that is dividing the surface reflection process from object into two kinds, namely the superficial reflection and the body reflection. The superficial reflection component generally has the same spectrum distribution with the incident light, the body reflects the component the spectrum to have the illumination spectrum power distribution and the object surface echoing characteristic decides together. The spectral component of the body reflection is decided by the power distribution of incident light and reflection characteristic of the object surface together.the body reflection has represented the object color echoing characteristic, may attribute the object color the Eigen characteristic. The model improvement based on the color ratio color invariant is made through the new construction color model, the superficial reflection component is canceled, the component of the body reflection only obtained, thus body index of reflection irrelevant to illumination and the geometry characteristic may obtained. Namely invariant information q1, q2, q3 correlated only with object color echoing characteristic, through the experiment result analysis compared with other color model (for example r, g, b model, HIS model and so on), the color model in this paper proposed has certain stability enhancement under the different light source as well as the different illumination strength.3. The color histogram reflection is the color statistic information overall, and the color spatial distribution information is lost. In order to increases the spatial information based on the color histogram method, this article proposed to process the fuzzy color histogram using the image piecemeal as well as the disposition weight, conforms to the human body visual rule, it can simulate the human's visual process to a certain extant.In view of the color histogram technology shortcoming, this paper proposed and implemented a new color-recognization method based on the Shafer reflection model and the fuzzy color histogram. Based on this color histogram method, the illumination stability is enhanced and the spatial information is increased. It can be used to simulate human's visual process, and has provided a new way to the color image processing technological development.