Fusion Model Of Visual Perception And Psychological Color Image Retrieval Technology Research

Multimedia technologies' popularizations and the Internet technology's implementation caused the massive pictorial information appearance, the tradition already not to be able to adapt the pictorial information retrieval request based on the text key word retrieval method, this enabled based on the content image retrieval technology to become the present research hot spot gradually.This article first introduced the topic significance and the background, to have made the simple analysis based on the content image retrieval technology's research present situation, based on this and discussed some key technologies which this domain involves. In the current practical application, often aims at the specific environment based on the content image retrieval system's design, and uses the specific algorithm. But the color characteristic is in the image physical feature the most direct visual characteristic, is opposite in other characteristics, the color characteristic is stable, regarding changes and so on image translation, criterion, revolving is insensitive, has the very strong robustness, moreover the color characteristic computation is simple. Therefore, in the image retrieval system which in the present paper realizes, mainly has used based on the color characteristic image retrieval algorithm. In the deep research image color feature extraction's foundation, This article constructs the chromaticity vector triangle in the introduction related color attention function's foundation in the HSV color space, and withdraws the chromaticity vector triangle interior contact radius R to do for the corresponding color space in the two-dimensional vector characteristic, and V carries on the quantification to its three dimensional characteristic brightness, union color spatial information extraction union matrix VR, namely the spatial color histogram does for the image retrieval characteristic, like this reduces the color space's three dimensional vector for two-dimensional, is away from using Euclidean space carries on the similar measure, through does with the traditional spatial color histogram for the characteristic retrieval compares. The experiment proved that this article proposes space color histogram image retrieval method can retrieve the picture target effectively based on HSV the color space VR from the image database.Subsequently, this article continues in the RGB color space the application geometry knowledge to each picture element R, G, the B picture element value structure colored vector triangular pyramid model, and withdraws the triangular pyramid the interior contact radius to take this picture element characteristic, like the original color image will transform as the gradation image, color space three dimensional transformation for univariate, the image information content reduces for original 1/3, similarly the union color's spatial information extraction gradation image's spatial color histogram does for the characteristic, uses the western-style distance to carry on the retrieval to the image. The experiment proved that this article proposes and the retrieval method can retrieve the picture target effectively based on the colored vector triangular pyramid interior contact radius gamma controller from the image database.Finally, this article to and realized based on the content image retrieval technology's research has carried on the summary; the difficulty which faced based on the content image retrieval technology has carried on the analysis to at present, proposed further studied direction.The main innovation lies in this article: [1] Proposed one kind based on HSV the color space VR space color histogram feature extraction method. This method had considered reasonably to a certain extent various colors component's relevance, transformed vector space's computation by the natural way the scalar computation, was one kind is different in the thought with the other algorithm new algorithm. The experiment proved that this method can retrieve the picture target effectively from the image database.[2] Further application geometry knowledge proposed one kind of gradation image transformation and the retrieval method. This method transformed similarly vector space's computation by the natural way the scalar computation, the experiment had indicated that this method was reasonable, is practical and feasible.