Color Matching Of Colored Fiber Blends Algorithm Based On Deep Learning

The past few decades have witnessed significant progress in artificial intelligence,especially in recent years;deep learning has brought advancements in many research areas.All walks of life have begun to march into artificial intelligence and seize the opportunities of the future.The computer color matching algorithm in the textile industry can best fit the torrent of computer development,and can be used as the information technology leader in the textile industry.Therefore,this thesis applies deep learning to the color matching algorithm of colored fibers,hoping to achieve the predicted ratio and the true value,improve the color matching efficiency,save labor costs,and achieve industrial upgrading.In this paper,the precolored cotton fibers were dyed using reactive dyes and mixed by a carding machine.The reflectance values of the samples were measured by Datacolor 650,and the raw data was sorted into a table as training data for the deep learning model.The TensorFlow toolkit Ludwig was then used to study the numerical relationship between the reflectance and the color ratio of the colored fibers.The sample data is organized into a CSV file,and the logical relationships of the construction are organized into YAML files,and the model training,verification,and prediction are performed by Ludwig.Finally,the model prediction value is compared with the real value to obtain a color matching result.Since Tensorflow was unable to find the logical value between reflectance and ratio,the color matching results were not satisfactory.MATLAB(2016A)was used to construct a deep neural model for color matching of precolored fibers.Two deep learning models P1 and P2 are constructed.There are 5 outputs in P1 model and 8 outputs in P2 model.Put the sample data into the model for deep learning and get thepredicted results.The sample data is then processed through the Stearns-Noechel model,and the processed data is placed into the model for deep learning.Compare the depth learning model to the accuracy of the predicted values of different sample data.The results show that the predicted value of the original sample data has the highest accuracy.It can be seen from the model regression curve of deep learning that the test curve is prone to a low R value.And the accuracy of the P1 model is always higher than the accuracy of the P2 model,Indicates that too many outputs will affect the result.By comparing the different prediction results,select the prediction result of the raw data of P1 model and perform color difference calculation to verify the color matching accuracy.The results show that the average color difference is 0.8431ΔE*ab,it is proved that the accuracy of depth calculation color matching algorithm basically meets the requirements.This paper shows that the deep learning model can be applied to the color matching algorithm of colored fibers,which can obtain more accurate color matching prediction values,which has important significance and application value for computer color matching technology.