Repeat Pattern Primitive Segmentation And Retrieval Algorithm Research Of Printed Fabric

The pattern reflecting the appearance of the fabric and giving the consumer the most direct visual impact,is one of the most important part of the fabric.It is one of the most important factors influencing people’s choice of fabrics and apparel.Cycling is an important feature of fabric printing patterns.Therefore,in the process of printing,the pattern primitive detection of repeat pattern is an important part of analyzing the fabric pattern.In addition,rich colors and various forms are another important feature of the fabric pattern.Printing companies design and produce a large number of patterns every day.With the accumulation of time,the traditional pattern retrieval method can not meet the needs of enterprises.Therefore,the storage and retrieval of a large number of patterns has become an urgent problem to be solved.With the development and application of computer image processing technology,fabric pattern recycling pattern primitive segmentation and pattern retrieval ushered in a new opportunity to use digital image processing technology to accurately segment the cyclic pattern and use it to map patterns.The fabric pattern quickly retrieves the same or similar patterns from the pattern library,overcomes the shortcomings of traditional manual segmentation and manual retrieval,and improves the efficiency of pattern design development.In this paper,the research on fabric cycle pattern primitive segmentation and cyclic pattern primitive retrieval is carried out.The research content mainly includes the following aspects.(1)A circular pattern primitive segmentation algorithm is proposed,which automatically determines a template image from the original image by using the edge density and the size of the image,and then uses the template image to match the original image for multiple targets,and searches for other regions matching the template,and then Through the displacement relationship between adjacent similar regions,a parallelogram is determined as the primitive size of the pattern,and finally the information loss formula is established to find a complete primitive pattern that conforms to the human visual perception.(2)Aiming at the characteristics of different patterns and isomorphisms of fabric patterns,a pattern retrieval algorithm based on edge and color features was proposed.After the median filtering and edge extraction,the three features of edge color histogram,edge direction histogram and edge distribution histogram are constructed.Then the Euclidean distance is used to measure the similarity,and the similarity is weighted and merged.Image similarity in the image library for image retrieval.(3)Aiming at the difficulty of retrieval caused by multiple segmentation results caused by inaccurate segmentation in the process of pattern element segmentation,an image matching algorithm based on SIFT feature extracted by cylindrical model is proposed.The primitive model is used to construct a cylindrical model,and each different primitive is transformed into the same cylindrical model,so that the pixel points are continuous up and down,and the left and right are connected.Then,based on the constructed cylindrical model,the SIFT features are extracted,and finally the SIFT feature points of different primitive images are matched.The experimental results show that the feature matching is achieved by the method proposed in this paper,and the matching degree of the image is significantly improved.