Three-dimensional Contour Measurement Based On Speckle Projection Stereo Vision

With the development of information technology,three-dimensional contour measurement is playing an increasingly important role in industrial detection,quality control,driverless,medical diagnosis and other fields.As an important way to realize 3D contour measurement,stereo vision has been widely concerned by researchers at home and abroad due to its non-contact and high efficiency,and gradually integrated with digital image related,speckle projection and other technologies.According to the process,three-dimensional contour measurement of stereo vision can be summarized as four steps: camera acquisition and system calibration,image correction,stereo matching,and transformation of three-dimensional information.Stereo matching effect is an important factor affecting the measurement accuracy.How to ensure good accuracy and speed under the influence of environmental noise and object surface characteristics is still an important research topic.In this paper,we combine the advantages of analog speckle projection and three-dimensional digital image correlation method,and improve the texture characteristics of the object surface through speckle projection.According to the process of speckle projection stereo vision,the generation and evaluation of speckle image simulation,the method of eliminating parallax error matching,the speed and sub-pixel accuracy of stereo matching algorithm are studied.This paper analyzes the influence of speckle distribution randomness and speckle size on speckle image quality,discusses the advantages and disadvantages of different stereo matching algorithms,uses zero mean normalized cross correlation and statistical transformation as the whole pixel search method,and optimizes the algorithm speed.The multi-directional scan line is used to aggregate the parallax cost,and the mismatched area is judged and filled by uniqueness detection and small area detection.The sub-pixel measurement is carried out by combining parabola method and Gauss Newton fastest descent inverse combination algorithm.In order to verify the feasibility of the algorithm and improve the accuracy and speed of three-dimensional contour measurement,an experimental platform of three-dimensional contour measurement based on speckle projection stereo vision is built in this paper.First,the speckle pattern is used to simulate and measure the displacement,and the maximum error of the average displacement is less than 0.01 pixel.In this paper,the three-dimensional surface contour of different moulds is reconstructed,and compared with the size of the experimental object,the measurement accuracy is 98.98%.