Yarn dyed fabric texture type recognition method

Abstract

The invention provides a yarn dyed fabric texture type recognition method, which comprises three steps of building of a double-sided gray level co-occurrence matrix, extraction of texture feature parameters and building of a BP neural network model. Through the above three steps, the front image and the back image of the yarn dyed fabric are converted into gray level images, and in a condition with a certain direction and distance apart, the joint probability for appearing of pixel pairs formed by pixel points on the front image and pixel points on the back image is calculated; later, contrast, correlation, entropy and homogeneity are selected as parameters for describing the image textures, and numerical values are calculated; and the extracted four texture feature parameters serve as input vectors, a texture structure type serves as an output vector, and a three-layer BP neural network is built for recognizing the texture structure type of the yarn dyed fabric. Through training of a certain amount of samples, the yarn dyed fabric texture structure type recognition accuracy is thus enhanced.

Description

technical field [0001] The invention relates to the technical field of automatic identification, in particular to a texture type identification method of colored fabrics. Background technique [0002] As an important feature of an image, texture can map the fine structure of a three-dimensional object surface to a two-dimensional image. It is widely used in the field of image classification and retrieval, but in many cases, grayscale texture cannot provide enough information. , so it is necessary to add color information to describe the object comprehensively and accurately. Color texture is a combination of image texture information and color information. It is a spatial color structure, which can be regarded as a color and structure distribution mode. [0003] In the field of texture structure recognition of woven fabrics, researchers have proposed many methods based on image processing technology to quickly detect the tissue structure parameters of woven fabrics. Among t...

AI Technical Summary

Problems solved by technology

However, most of the above methods can only be applied to white fabrics or monochrome fabrics, not only did not take into account the influence of the color information of colored yarns, but also only analyzed the texture features based on single-sided images, and could not record the double-sided texture and color information of fabrics.

Images