Research On Image Complexity Based On Color And Texture Features

Image complexity evaluation technology is widely used in image enhancement,information security and other fields.Studying image complexity can understand the human visual perception process of images,and it can also promote the development of image complexity technology applications.The study of complexity by domestic and foreign scholars mainly focuses on the feature expression of the image.In the evaluation process,human visual experience is not taken into account,and the proposed method is not applicable to all types of images.In view of these existing problems,based on visual theory,this paper proposes a study of the overall complexity of images based on color and texture features based on natural and texture images.For natural images,the main factors influencing the complexity evaluation are the color features.According to this,the evaluation of image complexity based on color features is proposed.In this paper,the color entropy,the color distribution uniformity,the second-order moments of the HSV three channels,and the color space complexity,which affect the complexity of the visual system evaluation,are firstly extracted.They are composed of feature vectors and are used to support a large number of images using the support vector machine method.Training was carried out to establish an image complexity level classification and complexity value calculation model,and the correctness of the method was verified through experiments.For texture images,the main factor affecting the complexity evaluation is the thickness of the texture.In this paper,the image complexity method based on texture features is studied.Firstly,the relevant texture parameters affecting the complexity factor are extracted through the GLCM,then the texture clustering center of the complexity level is obtained through the method of fuzzy clustering,and finally the distance between the image texture feature and the cluster center according to the image texture to be evaluated.Determine the category of the complexity of the image,and also give an accurate calculation model of complexity.Experimental results show that the study of image complexity based on color and texture features proposed in this paper not only conforms to the evaluation of human eye vision,but also is closer to subjective evaluation than other evaluation methods,and the complexity of the different types of images can be judged scientifically.