Research On Data Acquisition System Of Machine Vision Garment Sample Characteristic

In this paper,a non-contact input system based on machine vision is proposed for the fast collection and storage of garment 2D sample feature data.The system can capture the two-dimensional sample image of clothing by mobile phone,upload it to PC software system,and then automatically complete the calibration,preprocessing,vectorization,data conversion and automatic recognition and classification.It can replace the manual input operation of the traditional digitizer,and significantly improve the work efficiency.The research results show that the measurement accuracy of the input system can meet the actual needs and has application value.The main contents of this paper are as follows:(1)This paper analyzes the main problems existing in the process of manual collection of two-dimensional clothing sample data,and puts forward that using mobile phones to obtain images instead of traditional industrial cameras makes the system operation more flexible and convenient,and makes the obtained image range larger and accuracy higher.(2)The system automatic calibration method based on virtual grid technology is studied and implemented.Compared with the traditional two-dimensional plane calibration,virtual grid can effectively improve the calibration accuracy of the system.(3)Using image preprocessing,edge detection and other related technologies to extract the image edge contour is conducive to the subsequent feature extraction.(4)The corner detection technology is used to extract the edge corner points of two-dimensional clothing samples,and the edge segmentation is realized according to the corner points.At the same time,the straight and curved segments are classified and fitted.According to the requirements of the fitting accuracy of the straight segment and curve segment of the two-dimensional clothing sample,the fitting accuracy of the complex curve is effectively improved by the way of multi segment fitting.(5)By using the neural network technology,we can identify and classify the shape feature differences between different pieces,and input them into the database,which can provide effective data preparation for the big data analysis of the future clothing industry.(6)By automatically converting the edge vectorization data of two-dimensional garment samples into DXF file format,the data docking with garment design CAD software can be realized,which can meet the needs of later cutting processing and re editing.(7)The integrated design of the hardware and software system is completed,and the visual human-computer interface is friendly and easy to operate.