深度图像预处理算法研究

doi:10.16180/j.cnki.issn1007-7820.2021.11.005

摘要/Abstract

摘要：

为了解决RGB-D相机获取的原始深度图像数据包含噪声和空洞的问题,文中提出一种基于彩色图像引导的深度图像滤波算法。该算法基于双边滤波原理,以彩色图像作为引导图,将彩色图像灰度化后的灰度像素变化权值与高斯滤波权值相乘得到滤波综合权值,然后利用高斯核函数的分离特性,通过快速高斯变换降低计算的复杂度,提高算法的运行速度。实验结果表明,该算法在对原始深度图像进行滤波后,可显著提高深度图去噪效果,在保持双边滤波器去噪保边优点的同时,能对深度图像的小面积空洞区域进行修复,缩短运行时间。

关键词: RGB-D, 深度图像, 噪声, 空洞, 双边滤波器, 快速高斯变换, 去噪保边

Abstract:

In order to solve the problem that the original depth image data acquired by RGB-D camera contains noise and holes, a depth image filtering algorithm based on color image guidance is proposed in this study. The proposed algorithm is based on the principle of bilateral filtering. First, the color image is used as a guide map, and the gray pixel change weight of the color image is multiplied by the Gaussian filter weight to obtain the comprehensive filter weight. Then, the algorithm uses the separation characteristic of the Gaussian kernel function to reduce the computational complexity and improve the running speed of the algorithm through fast Gaussian transformation. Experimental results show that after filtering the original depth image, the algorithm can significantly improve the denoising effect of depth image. Additionally, the algorithm can repair the small-area void area of the depth image while maintaining the advantages of bilateral filter denoising and edge preservation, and shorten the running time.

Key words: RGB-D, depth image, noise, holes, bilateral filtering, fast Gaussian transformation, de-noising and edge preserving

中图分类号:

TN911.73

吴海波,王晨,崔禹. 深度图像预处理算法研究[J]. 电子科技, 2021, 34(11): 31-36.

WU Haibo,WANG Chen,CUI Yu. Research on Depth Image Preprocessing Algorithm[J]. Electronic Science and Technology, 2021, 34(11): 31-36.

图/表 8

图1

图2

图3

表1

图4

图5

表2

表3

参考文献 19

[1]	付梦印, 吕宪伟, 刘彤, 等. 基于RGB-D数据的实时SLAM算法[J]. 机器人, 2015, 37(6):683-692.
	Fu Mengyin, Lü Xianwei, Liu Tong, et al. Real-time SLAM algorithm based on RGB-D data[J]. Robot, 2015, 37(6):683-692.
[2]	Teng C H, Chuo K Y, Hsieh C Y. Reconstructing three-dimensional models of objects using a kinect sensor[J]. The Visual Computer, 2018, 34(11):1507-1523. doi: 10.1007/s00371-017-1425-2
[3]	张志华, 刘政怡. 多角度融合的RGB-D显著检测[J]. 计算机工程与科学, 2018, 40(4):681-689.
	Zhang Zhihua, Liu Zhengyi. RGB-D saliency detection based on multiple perspectives fusion[J]. Computer Engineering and Science, 2018, 40(4):681-689.
[4]	Husian F, Schulz H, Dellen B, et al. Combiningsemanticand geometric features for object class segmentation of indoor scenes[J]. IEEE Robotics and Automation Letters, 2016, 2(1):49-55. doi: 10.1109/LRA.2016.2532927
[5]	郑欢. 基于Kinect的深度图像修复技术研究[D]. 西安:陕西师范大学, 2016.
	Zheng Huan. Research on depth image inpainting technology based on kinect[D].Xi'an: Shaanxi Normal University, 2016.
[6]	秦兴, 高晓琪, 陈滨. 基于压缩卷积神经网络的图像超分辨率算法[J]. 电子科技, 2020, 33(5):1-8.
	Qin Xing, Gao Xiaoqi, Chen Bin. Image super-resolution algorithm based on squeezenet convolution neural network[J]. Electronic Science and Technology, 2020, 33(5):1-8.
[7]	Daribo I, Tillier C, Pesquet-Popescu B. Distance dependent depth filtering in 3D warping for 3DTV[C]. Chania:Proceeding of the IEEE Ninth Workshop on Multimedia Signal Processing, 2007.
[8]	Feng S, Murray-Smith R, Ramsay A. Position stabilisation and lag reduction with gaussian processes in sensor fusion system for user performance improvement[J]. International Journal of Machine Learning & Cybernetics, 2017, 8(4):1167-1184.
[9]	雷超阳, 刘军华, 张敏. 一种基于自适应的新型中值滤波算法[J]. 计算机工程与应用, 2008, 44(12):60-62.
	Lei Chaoyang, Liu Junhua, Zhang Min. New median filter algorithm based on adaptive[J]. Computer Engineering and Applications, 2008, 44(12):60-62.
[10]	Newcombe R A, Izadi S, Hilliges O, et al. KinectFusion: Real-time dense surface mapping and tracking[C]. Piscataway:Proceeding of the Tenth IEEE International Symposium on Mixed & Augmented Reality, 2012.
[11]	Du H, Miao Z. Kinect depth maps preprocessing based on RGB-D data clustering and bilateral filtering[C]. Wuhan:Proceeding of the 2015 Chinese Automation Congress IEEE, 2015.
[12]	Lin Y. The design of image depth information extraction algorithm based on joint bilateral filtering[J]. Journal of Physics: Conference Series, 2019, 1187(4):1-10.
[13]	Khoshelham K, Elberink S O. Accuracy and resolution of kinect depth data for indoor mapping applications[J]. Sensors, 2012, 12(2):1437-1454. doi: 10.3390/s120201437 pmid: 22438718
[14]	李少敏, 张倩, 王沛, 等. 基于高斯混合模型的Kinect深度图像增强算法[J]. 上海师范大学学报(自然科学版), 2016, 45(1):28-33.
	Li Shaomin, Zhang Qian, Wang Pei, et al. Kinect sensor's depth image enhancement based on Gaussian mixture model[J]. Journal of Shanghai Normal University(Natural Science Edition), 2016, 45(1):28-33.
[15]	Zhang X L, Dai L Q. Fast bilateral filtering[J]. Electronics Letters, 2019, 55(5):258-260. doi: 10.1049/ell2.v55.5
[16]	Yoshizawa S, Belyaev A, Yokota H. Fast gauss bilateral filtering[J]. Computer Graphics Forum, 2010, 29(1):60-74. doi: 10.1111/cgf.2010.29.issue-1
[17]	Handa A, Whelan T, McDonald J, et al. A benchmark for RGB-D visual odometry,3D recons-truction and SLAM[C]. Hong Kong:Proceeding of the IEEE International Conference on Robotics and Automation, 2014.
[18]	周自顾, 曹杰, 郝群, 等. 保留边界特征的深度图像增强算法研究[J]. 应用光学, 2018, 39(2):200-206.
	Zhou Zigu, Cao Jie, Hao Qun, et al. Depth image enhancement algorithm for preserving boundary[J]. Journal of Applied Optics, 2018, 39(2):200-206.
[19]	王鸿闯, 胡晓辉, 李薇. 一种基于改进阈值函Contourlet域的图像去噪算法[J]. 电子科技, 2019, 32(4):44-48.
	Wang Hongchuang, Hu Xiaohui, Li Wei. An image denoising algorithm based on improved threshold function Contourlet domain[J]. Electronic Science and Technology, 2019, 32(4):44-48.

	原始深度图像Ⅰ	原始深度图像Ⅱ
均值滤波	21.397 4	21.337 9
中值滤波	22.259 4	22.228 3
双边滤波	40.323 4	40.102 3
本文算法	44.355 5	45.531 6

	原始深度图像Ⅰ	原始深度图像Ⅱ
均值滤波	17	20
中值滤波	56	68
双边滤波	904	1 132
本文算法	77	80