针对局部立体匹配算法对噪声与光照变化敏感及在弱纹理区域匹配效果不佳的情况,提出一种基于改进Census变换与自适应参数引导滤波的立体匹配算法。该算法融合HSV通道值计算代价,对噪声与光照变化有较好的鲁棒性;在支持窗口内分别使用欧式距离加权与颜色加权获得Census变换参考值;集成AD与Census代价提高了单像素匹配代价的稳定性,降低了原Census变换对中心像素的依赖程度。在代价聚合过程中使用峰度系数对正则化参数进行自适应处理,通过视差计算获得视差图。在VS2017平台上对Middlebury数据库提供的图像对进行匹配实验,该算法在标准图像、加噪声图像及光照变化图像的各区域平均误匹配率分别是7.80%、10.72%和9.89%。结果表明:该算法可以降低误匹配率,同时能更好地适应噪声与光照变化。
Abstract
Aiming at the problems of the noise and illumination sensitive and with high false matching radio in the weak texture region of the local matching algorithm, an robust stereo matching algorithm for improved Census transform and adaptive parameter-guided filter is proposed. The hue, saturation, value(HSV) channel value is used to calculate the cost, which has higher robustness to noise and illumination changes. In the support window, the distance weighting and color weighting are used to obtain the reference value of the census transform respectively. The two costs are fused to improve the stability of the single pixel matching cost. The kurtosis adaptive regularization parameter is used on the guide filter for cost aggregation. Finally, the disparity map is obtained by disparity calculation. The matching experiment is carried out on the image pairs provided by Middlebury database on the VS2017. The average error matching rate of the proposed algorithm of standard image, noisy image and illumination change image is 7.80%, 10.72% and 9.89% respectively. The results show that the algorithm can reduce the false matching ratio and better adapt to the changes of noise and illumination.
关键词
计量学;立体匹配;图像处理 /
改进Census变换;代价计算;自适应参数;误匹配率
Key words
metrology /
stereo matching /
image processing /
improved census transform /
cost compute;adaptive parameter /
false matching ratio
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
［1］李鹤喜, 李威龙. 基于可变形卷积的双目视觉三维重建［J］. 计量学报, 2022, 43(6): 737-745.
Li H X, Li W L. Binocular Vision 3D Reconstruction Based on Deformable Convolution. Acta Metrologica Sinica［J］, 2022, 43(6): 737-745.
［2］张远辉, 徐栢锐, 朱俊江, 等. 可见光与热像融合的三维温度模型重建［J］. 计量学报, 2022, 43(2): 256-263.
Zhang Y H, Xu B R, Zhu J J, et al. 3D Temperature Model Reconstruction Based on Fusion of Visible and Thermal Images. Acta Metrologica Sinica［J］, 2022, 43(2): 256-263.
［3］王嘉明, 乔卫东, 张恒, 等. 基于阈值增量的动态眩光测量方法［J］. 计量学报, 2022, 43(4): 475-481.
Wang J M, Qiao W D, Zhang H, et al. Dynamic Glare Measurement Method Based on Threshold Increment. Acta Metrologica Sinica［J］, 2022, 43(4): 475-481.
［4］Scharstein D, Szeliski R. A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence Algorithms［J］. International Journal of Computer Vision, 2002, 47(1): 7-42.
［5］Sun J, Zheng N N, Shum H Y. Stereo matching using belief propagation［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25(7): 787-800.
［6］Besse F, Rother C, Fitzgibbon A, et al. PMBP: PatchMatch Belief Propagation for Correspondence Field Estimation［J］. International Journal of Computer Vision, 2014, 110(1): 2-13.
［7］Worby J, Maclean W J. Establishing Visual Correspondence from Multi-Resolution Graph Cuts for Stereo-Motion［C］//IEEE Fourth Canadian Conference on Computer and Robot Vision. Montreal, Canada, 2007.
［8］Kanade T, Okutomi M. A stereo matching algorithm with an adaptive window: theory and experiment［J］. IEEE Transactions on Pattern Analysis Machine Intelligence, 1994, 16(9): 920-932.
［9］Liu W, Anguelov D, Erhan D, et al. Ssd: Single shot multibox detector［C］//European conference on computer vision. Amsterdam, Netherlands, 2016: 21-37.
［10］Hirschmüller H, Innocent P R, Garibaldi J. Real-Time Correlation-Based Stereo Vision with Reduced Border Errors［J］. International Journal of Computer Vision, 2002, 47(1): 229-246.
［11］Zabih R, Woodfill J. Non-parametric local transforms for computing visual correspondence［C］//European Conference on Computer Vision. Berlin, Germany, 1994.
［12］Mei X, Sun X, Zhou M, et al. On building an accurate stereo matching system on graphics hardware［C］//IEEE International Conference on Computer Vision Workshops. Barcelona, Espana, 2011: 467-474.
［13］李海滨, 董玉, 李雅倩. 基于自适应参数的匹配代价融合算法［J］. 计量学报, 2017, 38(3): 280-283.
Li H B, Dong Y, Li Y Q. A Combined Cost Algorithm Based on Adaptive Parameter［J］. Acta Metrologica Sinica , 2017, 38(3): 280-283.
［14］Yoon K J, KweonI S. Adaptive support-weight approach for correspondence search［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006, 28(4): 650-656.
［15］He K, Sun J, Tang X. Guided image filtering［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(6): 1397-1409.
［16］Pauline T, Pascal M. Stereo Disparity through Cost Aggregation with Guided Filter［J］. Image Processing On Line, 2014, 4: 252-275.
［17］Chang Y C, Tsai T H, Hsu B H, et al. Algorithm and Architecture of Disparity Estimation With Mini-Census Adaptive Support Weight［J］. IEEE Transactions on Circuits and Systems for Video Technology, 2010, 20(6): 792-805.
［18］范海瑞, 杨帆, 潘旭冉, 等. 一种改进Census变换与梯度融合的立体匹配算法［J］. 光学学报, 2018, 38(2): 260-27.
Fan H R, Yang F, Pan X R, et al. Stereo Matching Algorithm for Improved Census Transform and Gradient Fusion［J］. Acta Optica Sinica,2018, 38(2): 260-270.
［19］方红萍, 曾若云, 吴家鑫, 等. 基于峰度系数方差的加权自适应引导滤波［J］. 北京理工大学学报, 2021, 41(11): 1193-1200.
Fang H P, Zeng R Y, Wu J X, et al. Weighted Adaptive Guided Filtering Based on Kurtosis Variance［J］. Transactions of Beijing Institute of Technology, 2021, 41(11): 1193-1200.
基金
国家自然科学基金(61573249);辽宁省自然科学基金(2019-MS-246);辽宁省教育厅基金(LZGD2019002);辽宁省高等学校创新人才项目(LR2019048);辽宁省研究生教育教学改革研究资助项目(LNYJG2022073);沈阳工业大学重点科研基金(ZDZRGD2020004)
PDF(72315 KB)
