Reaserch On Repeated Texture Elements Extraction From Texture Images

Element extraction method for texture image is refer to extracting the data structure composed of repeating unit(i.e.,a texture element)which has same or similar characteristics.Texture element extraction is in order to distinguish the area with similar visual features in a texture image,simplify or change the representation of the image and return it to a repeated combination of simple and independent texture element.It can make the image more easily understood and analyzed by computer,as well as provide the necessary precondition for the design of texture image.All of these divided regions have the following properties: Some areas represent the basic structure of the main texture elements;While others are homogeneous regions of similar color,that is,the texture background region.The project was supported by the National Natural Science Foundation of China Youth Science Foundation Project “The Natural Science Foundation of China under Grant”(No.61402053).On the one hand,the extraction of texture elements can decompose the texture image into an independent module,and then effectively utilize the extracted texture elements to analyze the topological structure of the original texture image.In addition,the individual texture elements represent the basic structure of the texture image,which is conducive to the study of texture image composition.On the other hand,the extracted texture elements can also be used for texture synthesis or new texture image generation.It can provide a strong prerequisite for the transfer,design and combination of the texture image.In order to cut out the repeated texture elements in the texture image,this paper presents an interactive element extraction method for texture image,which is able to simultaneously cut out the repeated texture elements with similar color or texture features using much less user interaction.A large number of experiments fully verify the practicability and effectiveness of the algorithm in this paper.The algorithm in this paper is mainly divided into following steps:Firstly,texture image can be segmented into individual and connected sub-patches by using Mean Shift algorithm,and their connectivity can be constructed by using Delaunaytriangulation.Secondly,we redefine a robust measurement distance by considering texture feature and optimized color feature measurement method to achieve high quality similarity judgment for texture elements with more obvious texture feature;Finally,by combining the new measurement distance,we further modify the optimization model,so as to achieve the best texture extraction.Begins from the comprehensive appearance,we set up a more effective and comprehensive measurement,which can synthesize the corresponding features of the texture elements thus avoid the limitations of an algorithm only be effective for some specific appearance characteristics.Comparing a large number of experimental data found that the proposed method gets a great better extraction result for the texture image that has foreground and background with similar color,and improves the speed of image segmentation significantly.