Research On 3D Structured-light Measurement Technique For Industry Scene Measurement

3D structured-light measurement technique is widely applied in the industry detection field, such as reverse engineering, intelligent navigation and pose measurement and matching. On-site measurement has requirement for high speed, real-time capability and good robustness against complicated illumination condition. Surface structured light projection could satisfy the real time request with high measuring effectivity and the ability. Among different kinds of encoded structured light patterns, sinusoidal grating patterns could get result of highest resolution. The major error source of sinusoidal grating patterns is the nonlinear distortion of projector. Most of the existing correction methods have problems of relying heavily on gamma distortion model and large calculating quantity. Techniques using article parallel light without coded information have good adaptability of complicated illumination and anti-noise performance, but it is hard to match corresponding uncoded stripes.Directing against the problems of existing correction methods for nonlinear distortion, a new method based on LUT is proposed in the paper, which is focused on correcting the input gray and with the goal of improving universality and reducing the complexity. There is no need to calibrate the mathematical model of input-output characteristic curve. By computing the input gray corresponding to each sampling point of the sinusoidal output waveform, A input-output LUT is built up. Then the output waveform is corrected by the correction of input. This kind of method is of simple steps and small calculating quantity. The result of phase measurement experiment shows it could restrain the content of harmonious efficiently.Because of the harsh on-site industrial environment,the measuring system is expected to be not sensitive to ambient light as well as with a good anti-noise performance in order to measure the workpiece on site, the surface of which is rough and has low reflectivity. Line structure light use the centerlines of light stripes as feature points, having good robustness with the gray of single pixels. For this reason, a 3D shape measurement system based on line structure light is designed as research platform. The result of measurement of targets with rough surface correctly reflect the geometric feature of the target, the detection precision achieve 0.25 mm.In order to match corresponding points of parallel light,a new matching algorithm based on epipolar geometry constraints as well as symmetry validation is proposed. In the new algorithm,the structural parameters of the system are used to convert the light-plane-equation between two camera coordinate systems,as for each virtual matching point of the light stripes’ center points of on left-image epipolar, by computing their left-image corresponding point and comparing the distances of reverse result and origin points,the light stripes in left image and right image could be matched correctly.Twenty light planes are used to simulate parallel light strips in the measurement experiment, the result of measurement correctly reflect the geometric feature of the target, which shows the new algorithm is reliable.