Fuzzy Image Target Recognition Algorithm Based on Visual Similarity in the Internet of Things Environment

Abstract

Abstract Because of the low accuracy of fuzzy image target recognition, an algorithm of fuzzy image target recognition based on visual similarity was researched. First, the method of adaptive weighted mean threshold was used to deal with fuzzy images, and then the adaptive threshold was obtained by a full scan of a fuzzy image. Second, the pixel point enhancement equation was established. Moreover, the new gradient operator used the points after filtering enhancement to reconstruct pixels and obtain the deblurred image. In addition, the visual similarity method was used to extract target features of the deblurred image, and then the color image was converted to a relatively uniform color space. On the basis of visual spatial response characteristics, brightness, contrast function, and chroma were used to adjust the image and thus to obtain the structure similarity index of each dimension image. By comprehensively considering the information of each dimension in color space, the structure similarity index was used to extract the image target features. Finally, the support vector machine model learned the target feature samples. Experimental results show that the proposed algorithm can effectively identify the fuzzy image target.