Stereo Matching Algorithm for Weak-Texture Region of Glass-Ceramic Based on Improved Local Entropy

Ying Li; Lu Dai; Jiaqi Wang; Jinkai Xu

doi:10.3788/LOP250570

Laser & Optoelectronics Progress, Volume. 62, Issue 16, 1616001(2025)

Stereo Matching Algorithm for Weak-Texture Region of Glass-Ceramic Based on Improved Local Entropy

Ying Li^1,2、*, Lu Dai^1,2, Jiaqi Wang^2、**, and Jinkai Xu²

Author Affiliations

¹School of Optoelectronic Engineering, Changchun University of Science and Technology, Changchun 130022, Jilin , China

²Key Laboratory for Cross-Scale Micro and Nano Manufacturing, Ministry of Education, Changchun University of Science and Technology, Changchun 130022, Jilin , China

show less

Abstract Get PDF(in Chinese)

Figures & Tables(11)

Fig. 1. Flowchart of SIFT feature matching algorithm based on local entropy

Download full size

Fig. 2. Flowchart of matching algorithm

Download full size

Fig. 3. Experimental devices. (a) Schematic diagram of devices; (b) physical map of devices

Download full size

Fig. 4. Disparity maps obtained by different algorithms. (a) Original images; (b) ground truth disparity maps; (c) SGM; (d) AD-Census; (e) PatchMatch; (f) proposed method

Download full size

Fig. 5. Comparisons of matching rates for different algorithms. (a) Comparison of matching rates for different algorithms on four datasets; (b) comparison of mismatch rates for non-occluded areas on four datasets

Download full size

Fig. 6. Binocular images of glass-ceramic. (a) Image captured by CCD 1; (b) image captured by CCD 2

Download full size

Fig. 7. Matching results. (a) SIFT feature matching results without local entropy; (b) local entropy images; (c) SIFT feature matching results with local entropy; (d) sparse disparity result; (e) matching result

Download full size

Table 1. Matching rates of four algorithms on different datasets
View table
Table 1. Matching rates of four algorithms on different datasets
Algorithm Teddy Cones Tsukuba Venus Average
SGM 85.35 85.74 85.52 81.62 84.56
AD-Census 86.45 87.15 87.31 82.16 85.77
PatchMatch 88.89 90.01 88.43 75.45 85.70
Proposed algorithm 91.74 92.82 91.93 89.32 91.45

Table 2. Mismatch rates of four algorithms for non-occluded areas on different datasets
View table
Table 2. Mismatch rates of four algorithms for non-occluded areas on different datasets
Algorithm Teddy Cones Tsukuba Venus Average
SGM 9.43 6.14 10.08 13.63 9.82
AD-Census 5.74 6.54 8.62 11.27 8.04
PatchMatch 10.06 7.01 9.21 11.57 9.46
Proposed algorithm 5.33 3.96 6.12 8.70 6.03

Table 3. Comparison of feature matching results
View table
Table 3. Comparison of feature matching results
Algorithm Number of initial matching points Number of matching points after screening Matching rate /%
SIFT 202 118 58.42
SIFT + local entropy 531 495 93.22

Table 4. Error values of four algorithms
View table
Table 4. Error values of four algorithms
Algorithm RMSE MAE
SGM 25.19 24.56
AD-Census 23.42 22.83
PatchMatch 21.72 20.24
Proposed algorithm 18.67 18.12

Tools

Get Citation

Copy Citation Text

Ying Li, Lu Dai, Jiaqi Wang, Jinkai Xu. Stereo Matching Algorithm for Weak-Texture Region of Glass-Ceramic Based on Improved Local Entropy[J]. Laser & Optoelectronics Progress, 2025, 62(16): 1616001

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites