Three-Dimensional Surface Measurement Of Objects Based On Multi-Line Structured Light

Three-dimensional(3D)measurement technology has a very broad range of applications,such as industrial measurement,heritage conservation,etc.One of the common 3D measurement methods is the line structured light 3D vision measurement technique.This method generally uses a single-line laser in combination with stepper motors,slide rails,rotating stages and other motion devices.By projecting the laser onto the surface of the object to be measured and combining the device parameters with the camera parameters,the 3D measurement of the object can be completed.However,the sliding accuracy of the slide rail,the rotation accuracy of the rotary stage and the displacement accuracy of the stepper motor will all have an impact on the final measurement results.Therefore,this paper proposes a new multi-line structured light measurement method based on the principle of 3D measurement of line structured light,which discards the precision mobile equipment required in the traditional method,and only combines single-line structured light with three-line structured light to complete the 3D measurement of the object,thus reducing the error caused by the equipment.In this method,the single-line structured light is projected onto the object surface,the three-line structured light remains fixed,and the single-line structured light moves and intersects with the three-line structured light to form three intersection points,and by using the camera coordinates of these three intersection points,the real-time calibration of the scanned light plane can be completed.In this process a single-line structured light can be scanned at any angle to determine the complete 3D shape of the object.In addition,due to lighting conditions and factors such as errors in experimental instruments or human interference,the point cloud obtained from the 3D measurement method based on structured light often contains noise,which can result in rough or deformed reconstructed object models.Therefore,this paper proposes a point cloud denoising method that combines an improved DBSCAN algorithm and an improved K-nearest neighbor algorithm.For the first time,the rate of change of normal vector angles is introduced into the Gaussian kernel function,which effectively removes point cloud noise while preserving the fine details of the objects.The effectiveness and accuracy of the proposed method are further validated through denoising experiments on open-source point clouds.The experimental results indicate that the proposed method overcomes the difficulties in obtaining certain angle and position information,and effectively reconstructs the three-dimensional shape of the object.The repeatability error of the system measurements is lower than 0.16 millimeters,which is sufficient for measuring the three-dimensional morphology of complex workpieces.