Research On Bilateral Filtering Denoising And Watershed Segmentation Of Point Cloud Depth Image

In recent years,due to the good characteristics of depth images and their wide application in the fields of human-computer interaction and cultural heritage protection,a large number of experts and scholars have carried out research on depth images.However,high-precision,precise and clear high-quality depth image acquisition equipment is expensive,so the use of point clouds to generate depth images has become a common method.However,when the point cloud data is generated as a depth image,some image noise will be generated.If the image contains noise in the subsequent processing,it will cause loss of image details,resulting in poor subsequent processing effects.Therefore,before performing image processing on the point cloud depth image,the depth image must be denoised.And the point cloud depth image generated by the point cloud data can be processed by image processing methods,such as segmentation,registration,point cloud boundary extraction,etc.,to provide new ideas for point cloud data processing.In response to the above problems,the main research content and research results obtained in this paper are as follows:1)Aiming at the problem that image noise will be generated when the point cloud data is generated as a depth image,which will affect the subsequent processing,this paper uses filtering and denoising methods to denoise the depth image.In order to explore which filtering method has better effect in removing noise and keeping edges,this paper selects four common filtering methods in image processing for comparison.Through principle analysis,image analysis of detection results,and edge preservation index analysis for comprehensive comparison and analysis,we can know the bilateral filtering algorithm.The effect is better in point cloud depth image denoising and edge preservation.2)In view of the image noise generated when the point cloud data is generated into the depth image,analyze the type of noise generated and the nature of the bilateral filtering algorithm,and improve the bilateral filtering algorithm in a targeted manner.First judge whether the center of the template is a salt and pepper noise point,if so,use the scalable median filter algorithm to denoise,and then use the bilateral filtering algorithm improved by gray-scale similarity to perform secondary denoising processing on the template after median filtering..In this paper,by adding noise to the ordinary image and then denoising processing and point cloud generation depth image,performing denoising processing for many experiments to prove the effectiveness of the improved bilateral filtering algorithm in this paper.The results show that the improved algorithm has better denoising effect and has better effect on the edge.Better protection.3)According to the current research status of point cloud data segmentation and the application of watershed algorithm in image processing,a point cloud segmentation method based on point cloud depth image and watershed algorithm is constructed.After generating point cloud depth image from point cloud data and establishing index,bilateral filtering algorithm is used for filtering and denoising,watershed algorithm is used for image segmentation,and the segmentation results are indexed back to the original point cloud data to obtain the point cloud segmentation results.In order to verify the reliability and accuracy of the method,the region growth method,RANSAC algorithm and Euclidean clustering method are used for comparative experiments.Through the comparative analysis of the experimental results,this method can effectively segment different point cloud patches,and has good accuracy and integrity,high quality,which provides a new idea and method for point cloud data segmentation.