Calibration And Measurement Of Structured Light Vision Inspection System

As the development of modern industrial technology, three-dimensional measurement research technology is getting more focus on the field of measurement. Among them, the structured light visual inspection technology has the advantages of high precision, large range, extracting light bar information easily, especially for measuring the object with a deformable or curved surface.Visual inspection of structured light is non-contact. During the detection process, the laser will emit at the surface of the object. At the same time, the structured light stripe will change with the variation of the surface and depth, which produces distortion. The information of the measured surface depth and the relative position between the laser and the camera are contained in the two-dimensional image distorted, and the task of the structured light visual inspection is to obtain such information by extracting structured light strip center position. The detection of structured light strip center position is a key aspect of structured light vision inspection technology, so the accuracy of structured light strip center extraction will affect the performance of the entire structure of the optical detection system directly.Since the camera calibration is also the key technology in the entire three-dimensional structured light vision inspection system, the paper first studied the calibration, including a variety of camera calibration methods and principles of the light plane calibration method. Its advantages and disadvantages will be analyzed and discussed one by one in this paper.In a structured-light vision system, the accuracy of center position of the stripe is of great importance for the whole measuring system. According to the problem of the low accuracy of extracting the light stripe center position in the measuring system,a new method of extracting structured light strip center is proposed in this paper.A way of detecting the structured light strip center line is based on Gaussian distribution curve fitting method, which is considered the fact that light intensity of the stripes center presents a Gaussian distribution status approximately is proposed. In this fitting process, we firstly process the saturation point that influences the curve fitting. After that we should do the curve fitting again, so that the fitted Gaussian curve is closer to the ideal Gaussian curve, and then the location of the light strip center can be determined by confirming the center coordinate of the fitting curve. Experimental results show that:this improvement of the Gaussian distribution curve fitting method not only improves the accuracy of extracting light strip center, but also has better noise immunity compared to the previous method. The results verified that this method can efficiently extracted the center-line of light stripe, and this method has improved the accuracy of 3D reconstruction.