Study On Glass Defects Moire Image Real-time Processing Method Based On DSP

In glass production process, because of the limitation of float glass production process,in the production process,the defects not only damage the appearance and opticalhomogeneity of glass products, but also directly affect the glass product quality grades, theon-line detection of glass defects for improving the quality of glass products have importantpractical significance.In this paper, glass defect moire image processing can be realized by using DSP whichwas adopted to achieve the online glass defect detection method based on projection gridline to overcome the problem of a large amount of computation of the Fourier transformimage moire.Based on the characteristics of Fourier transform, the design of the hardware andsoftware of the glass defect detection system and the selection of the hardware wereprovided to realize the acquisition, processing, analysis and data communication of moireimage information.In this paper, the HPI interface of DSP was used to realize multiple DSP datacommunication with the host and the data transmission between the linear CCD and DSP. Inthe design of initialization software, On-chip peripherals including Flash, SDRAM anddouble-DMA were initialized to complete extracting the spectrum of moire by band-passfilter in order to get the image of phase main value.Unwrapping algorithm based on the phase jump correction was used to completeexpanding the phase difference and thresholding segmentation method based on the imageof phase true value difference was adopted to get the scope and size of the defect.It turns out that the detection accuracy of the method can reach to0.1mm to realize thereal-time detection of below800m/s, which could meet the requirements of the design.The study of this paper provided the foundation based on the online testing of the glass ribbon,which has a great practical meaning for improving glass production process and thequality of glass products.