Research On Defect Detection Method Of Glass Panel Based On Structured Light

Due to the complicated production process,glass panels are prone to various defects in the production process.Traditional human eye detection is increasingly difficult to meet the needs of manufacturers.In order to improve product quality and reduce production costs,businesses urgently need non-contact,automated,accurate,and highly efficient detection methods to replace the manual detect method.3D measurement technology based on structured light can meet this demand.This paper mainly studies the application of structured light three-dimensional measurement technology to defect detection of glass panels.In this paper,the principle analysis of the glass panel defect detection system of is analyzed,and the factor that influence the system's precision is studied.In detail,the image processing algorithm,random phase shift problem and the contrast of detection system are studied in this paper.1.In the image processing algorithm,this paper mainly studies the defect acquisition mode,mask,denoising,contrast enhancement and binarization.Specific content is as follows: analyzes the advantages and disadvantages of different defects representation,this article from defect characterization methods of noise robustness and the sensitivity of the two into consideration to decide choose what way for defect information,found that the system is influenced by noise is small,small changes over time,eventually selected modulation distribution of defect information;The advantages and disadvantages of making masks are analyzed,and the suitable mask is finally made.Proposed combines rough set multidirectional difference theory and wave theory is applied to image denoising,using rough set theory,according to the direction of the difference image and extremum direction of image pixels are classified,pixels can be divided into normal pixel set,pulse collection,random noise pixels.Then the normal pixel don't do processing,to deal with pulse pixels using adaptive pulse filter,then to curvelet denoising of images,and the simulation show that this method is better than single use curvelet denoising effect;Using gamma transformation,gray scale drawing and Laplace operator to enhance the image,there are several methods can improve image contrast effectively.then we use binaryzation algorithm to process the image.2.A method for calculating the modulation system of random phase shift isproposed.In this paper,a new method of adjusting system is proposed,the simulation and experimental demonstrate the effectiveness of the proposed method.3.A new method for improving fringe contrast was studied.The method is by introducing a lens to the PMD system to make the contrast of the fringe received on the object to be the highest.Experiments show that this method is effectively in improving the system contrast.