Color Card Sample Collection And Recognition System

In the spun dyed yarn,the fiber is dyed first,and then two or more kinds of fibers are mixed and spun in a certain ratio to form a fabric having a color difference.Because of its ability to realize various three-dimensional space effects and textures,the spun dyed yarn is deeply loved by consumers.At the same time,the production process of spun dyed yarn is low-carbon and environmentally friendly,with little pollution,broad development prospects and huge market space.However,at present,when spun dyed yarn are analyzed for their composition,color matching personnel use reverse engineering for color matching,and in this process,they are easily affected by various external factors.In addition,the use of artificial color matching is relatively inefficient,resulting in higher costs,which is a pain point and difficulty for enterprises.In order to solve the above problems,this thesis uses image recognition technology to research and develop the color sample collection and recognition system.Because the HSV color space is more in line with the human eye's perception,and the color it describes is easier to quantify,this paper chooses the HSV color space.Since the HSV color space value of the image cannot be directly obtained,it is first obtained by taking its RGB color space value and then converting it to HSV.In the thesis,the color histogram,the cumulative histogram,the color moment are used to describe the color features,and the data is expressed in the form of a histogram.The chi-square coefficient is used to calculate the distance between the histograms,and the distance is used to construct the features of the image.Function to measure the similarity of an image.Gaussian normalization is used to make each eigenvalue have the same position in the same space.Finally,the fusion of image color and texture features is performed.The texture features extracted by wavelet transform of Variable Tree-Wavelet Analysis and the features after fusion are measured by similarity using Euclidean distance.In this thesis,color features,texture features,and the fusion of the two features are used to the spun dyed yarn images,and the results are analyzed and compared.The results show that the image retrieval of the spun dyed yarn with color and texture features is higher and the robustness is better.According to the above conclusions,this thesis designs a color card sample collection and recognition system,which realizes the image retrieval function of the fusion feature algorithm.The system was tested and the results show that the system has high use value.