The Research Of Apparel Cutting Recognition Based On Vision

With the continuous development of science and technology,image processing technology and intelligent system equipment are constantly penetrated to more areas.Textile and garment manufacturing is a Labor-intensive industries,all aspects of its production process need to invest a lot of manpower and material resources,especially the clothing piece recognition process,workers are required to maintain a high concentration of mentality at all times,which has caused tremendous challenges to the workers' spirit and body.So,it is a trend to recognize the pieces with a reliable,intelligent and efficient system.The targe recognition system based on DSP microprocessor,not only is it convenient,flexible and portable,but also it has low power consumption and relatively low cost.Therefore,it has a very broad application prospect.This paper designs a clothing piece recognition system based on vision.First of all,we introduced on the image preprocessing method,the edge detection algorithm based on Canny and adaptive threshold segmentation localization algorithm,the feature extraction method based on Fourier shape descriptors and shape contours curve and the HSV color space model,the multi-class classification recognition algorithm of SVM,which lays a theoretical foundation for the recognition of this piece.Secondly,set up system hardware platform and configure software development environment,which including the installation and configuration of the camera,Qt library,and Open CV library.Finally,we transplanted the Open CV computer vision library and image user interface Qt into the DSP development panels,classification and recognition experiment for the algorithm presented in this paper,and the interface of Qt was used to complete the system's human-computer interaction.To improve the recognition rate of the system and reduce the running time of the algorithm,the features are extracted by combining the shape contour description with the HSV color model,this feature is then trained on the PC and transferred to the sample feature database in the DSP development board for classification and recognition.In order to verify the effectiveness of the algorithm,we compare the algorithm with other algorithms based on the created piece database,after many experiments and analysing,the results show that the proposed algorithm has greater advantages in recognition rate and recognition time,and achieved good results.