Evolutions and Key Techniques for 3D Object Surface Vision-Measurement in Industry

Digital close range photogrammetry system and machine vision are widely used in production control,quality inspection,and new production development in industry.The main aim is to provide precision 3D coordinates or reconstruction of an object surface and give an expression to an object shape.Current research findings indicate that calibration of digital camera has some problems in high precision measurement.The factors of causing image distortion have not been fully clear.Calibration model needs to be optimized for some application.Validity and adaptability of calibration need boosting.Detection,recognition and auto coding of artificial marks and coding marks and epipolar techniques are effective methods for coding of homonymous point in different images.Sub-pixel measurement precision is above 1/100 pixel under some condition.Our research focuse on automatic or semi-automatic measurement of object surface,and image match arithmetic.The arithmetic from single point matching to multi point matching and to multi photo matching is an important evolution. This paper reviews and analyzes systematically the key techniques for 3D object surface vision-measurement in industry,discusses and sums up new methods,new theory,new problem,and distortion parameter model of images,precision level in digital camera calibration.Several main artificial marks used in non-texture or poor-texture object surface measurement are introduced.The paper analyzes comparatively main methods for sub-pixel measurement and it can gain precision level in present,and indicates important evolutions in automatic and semi-automatic measurement,essential arithmetic or robust arithmetic for obtaining approximate coordinates of marks points in an object surface measurement.The paper also points out the importance of knowledge-dig and its application in mark recognition and detection,and emphasizes integrative techniques in vision-measurement system. In the end,a pilot study and a result of practice application are given.Research proved that the thought and methods about image segmentation,detection and location of circle marks were effective and valid. A propriety weight value for adding parameters,control points and oriented elements in self-calibration bundle adjustment are advantageous to gaining high precision of space coordinates.The R M S of check points of control point is ±0.2 mm.