Font Size: a A A

Research On Content Based Image Retrieval

Posted on:2014-04-03Degree:MasterType:Thesis
Country:ChinaCandidate:H ChenFull Text:PDF
GTID:2268330425968223Subject:Software engineering
Abstract/Summary:PDF Full Text Request
In recent years, with the rapid development of computer technology, informationexchange more frequent, digital image and video data and more and more of the imageretrieval technology, so more and more. Early method of text based image retrievaltechnique, but a larger workload, and subjectivity, resulting in retrieval effectiveness.Therefore, emerge as the times require use of content-based image retrieval technologybased on the image color, texture and shape features, image retrieval, this paper focuseson the combination of color, texture and method of retrieval algorithm and processimageRetrieval based on color feature, first of all, this chapter on the HSV and RGBcolor spaces are compared, experimental results show that, the HSV color space can bemore consistent with human visual judgment standard. Secondly, the paper select HSVcolor space, the validation method of choice etc., the selection of an appropriate colorspace, quantization standard requires the quantization of the color spaces, divided intotwo categories: one is the equal interval quantization and non-equal intervalquantization, extraction method by feature selection right, the general color histogramscan only express the global statistical information of image the spatial position, unableto express color, so this paper uses the method of cumulative histogram, and throughspecific test results will be compared with the cumulative histogram comparison. Finally,the similarity measure method selection, this paper compares the Euclidean distance andweighted distance, the retrieval results more effective.Another one of the core content of this paper is a retrieval method based on texture,introduces the analysis method of gray level co-occurrence matrix and tin based ontexture, this paper adopts the method of extracting the texture based on gray levelco-occurrence matrix, the texture feature parameter extraction method based on theimage, we use five kinds of texture parameters, respectively. Is: energy, entropy,correlation, partial equilibrium and moment of inertia, the use of these five parameters,image matching computing data distance, so as to realize the computer automaticretrieval.
Keywords/Search Tags:Image retrieval, Color histogram, Euclidean distance, TextureCharacteristics, GLCM
PDF Full Text Request
Related items