基于Kinect相机的深度图像修复算法  

Depth image restoration algorithm based on Kinect camera

作　　者：刘继忠[1] 王聪[1] LIU Jizhong;WANG Cong(Institue of Robotics,Nanchang University,Nanchang 330031,China)

机构地区：[1]南昌大学机器人研究所,江西南昌330031

出　　处：《南昌大学学报（工科版）》2021年第2期172-177,共6页Journal of Nanchang University(Engineering & Technology)

基　　金：国家自然科学基金资助项目(61863027,61273282);江西省重点研发计划资助项目(20202BBE53025,20202BBGL73057);南昌市双百人才计划资助项目。

摘　　要：针对Kinect相机获取深度图像存在空洞和噪声的问题,提出了一种基于联合双边滤波和中值滤波相结合的深度图像处理方法。将Kinect获得的彩色(RGB)图像和深度图像裁剪对齐,利用联合双边滤波器修复深度图像中的空洞,用中值滤波器滤除噪声。实验结果表明与其他方法相比,所提算法能够有效填充深度图中的空洞区域,并去除图像中的噪声,能够达到更好的修复效果。Aiming at the problem of holes and noise in the depth images acquired by the Kinect camera,a depth image processing method based on the combination of joint bilateral filtering and median filtering was proposed.First,the color(RGB)image obtained by Kinect and the depth image were cropped and aligned;then,the joint bilateral filter was used to repair the black hole in the depth image;finally,the median filter was used to filter out the noise.The experimental results showed that the proposed algorithm can effectively fill the black hole area in the depth map,remove the noise in the image,and achieve a better repair effect comparing with other methods.

关 键 词：KINECT 深度图像修复 联合双边滤波器 中值滤波器 噪声滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]