Weft Skew Detection Of Fabric Based On DSP+FPGA

With the continuous improvement of people’s living standards, the requirementsof textiles have become more and more demanding. In the textile production process,the skew is an important factor affecting the quality, skew detection has become thehotspot of the textile industry at home and abroad. In the country, photoelectric weftdevice is used to detect skew basically. Though this method is more labor-saving andmore accuracy than traditional human eyes, it has poor adaptability on thicker orintensive or complex weave woven twill and satin fabric, and is also difficult toenhance the accuracy of detection. In recent years, foreign countries have begun tostudy the camera weft, but the equipment is expensive, small and medium enterprisesin the domestic are unbearable.In order to solve this problem, this paper designed a camera weft detectionsystem based on DSP+FPGA structure. The system includes machine vision andimage processing technology, so it can be real-time detection and adjustment of weftoblique. The core of the system chip uses TI’s TMS320DM642, dedicated to theoperation of the algorithm with the data, the FPGA as auxiliary chip of the systemused to generate the control signals and simple pre-processing of image data collected.Camera Link technology for data transfer between the camera and the DSP, designedto meet the real-time requirements, in order to coordinate the data transmission andDSP receive rate in FPGA integrated structure of a ping-pong FIFO as data buffer.In connection with the textural properties of the fabric, two skew detectionalgorithm are introduced. One is skew detection algorithm based on Hough transformand the other is projection skew detection algorithm. In the line detection, althoughHough transform has higher fault tolerance and robustness, the amount ofcomputation and data storage is large, it can not meet the real-time requirements. Inthis article, Fourier transform and wavelet transform are used to extract the feature of projection skew detection algorithm,and the good results are satisfactory. For thissystem, projection skew detection algorithm is improved, when encounter the case ofsame or similar accurary, the algorithm can achieve real-time requirements of thesystem. Experimental results show that the algorithm has strong anti-jammingcapability on the fabric density, color and shape, its detection accuracy can reach95%,and can meet the requirements of real-time detection in the pipeline.