Recognition of Front Bodice Wrinkle Defects in Men's Suits Based on Image Processing
The poor front bodice wrinkles is one of the common drawbacks of suits,which is highly dependent on professionals,time-consuming and susceptible to subjective influence in the suit customization process.Taking men's suits as an example,this paper constructs an automatic recognition method for the defects of predecessor pleats.The male suit malpractice map was collected from the enterprise,and the target image was extracted by using the segmentation and anno-tation tool EISeg.The bicubic interpolation was used to unify the image resolution and standardize the feature parameters.The image was grayed,gamma transform and threshold segmentation to simplify the operation data,enhance the image information,and highlight the fold trend.According to the local gray curve of the fold,the width,depth and density of the fold are extracted.According to the threshold segmentation map,the fold direction and the fold position are extracted.The particle swarm optimization algorithm is added to the BP neural network to improve the network model and output the malpractice category.The results show that the test set accuracy of the optimized network model is 8.3%higher than that of the traditional BP neural network model.This method can accurately realize the automatic identification of the defects of the front of the men's suit,and bring new technical means and methods to the industry.
suit front defectimage processingparameter extractionparticle swarm optimizationBP neural network
万方数据
维普
