Author Affiliations
1School of Mathematics and Computer Science, Zhejiang A & F University, Hangzhou 311300, Zhejiang , China2Zhejiang Provincial Key Laboratory of Forestry Intelligent Monitoring and Information Technology, Hangzhou 311300, Zhejiang , China3Key Laboratory of State Forestry and Grassland Administration on Forestry Sensing Technology and Intelligent Engineering, Zhejiang A & F University, Hangzhou 311300, Zhejiang , Chinashow less
Fig. 1. Five species of tree
Fig. 2. Flowchart of tree disparity image generation method
Fig. 3. Median calculation
Fig. 4. Cost aggregation. (a) Minimum path cost; (b) 8-path aggregation
Fig. 5. Schematic of winner-take-all algorithm
Fig. 6. Correction results of tree image. (a) Before correction; (b) after correction
Fig. 7. Histogram equalization results before and after preprocessing. (a) Tree left image; (b) tree right image; (c) tree left image after preprocessing; (d) tree right image after preprocessing
Fig. 8. Disparity images generated by different algorithms. (a) SGBM algorithm; (b) BM algorithm; (c) SGM algorithm; (d) proposed algorithm
Fig. 9. Comparison between the disparity image generated by the proposed algorithm and the real disparity image. (a) Original left image; (b) real disparity image; (c) disparity image generated by the proposed algorithm
Parameter | Census size | P1 | P2 | r | Maximum parallax value | µ1 | µ2 | δd |
---|
Value | 3×3 | 10 | 150 | 8 | 64 | 0.8 | 1 | 1 |
|
Table 1. Algorithm parameters in this paper
Camera parameter | Left camera | Right camera |
---|
Focal length(fx,fy) | (1139.78060,1140.06362) | (1139.24564,1139.93787) | Optical center position coordinate | (679.74223,377.37711) | (683.14170,380.91245) | Distortion coefficient | (0.08655,-0.22816,0.00233,-0.00617,0) | (0.08655,-0.22816,0.00223,-0.00617,0) | R | [-0.01300,-0.00469,-0.00061] | T | [-60.73141,0.16710,0.42228] |
|
Table 2. Calibration results of binocular camera
Algorithm | False match rate /% | Time /s |
---|
Traditional Census | 14.19 | 4.7 | Improve Census | 11.23 | 4.9 |
|
Table 3. Matching cost test result
Algorithm | Tuskuba | Teddy | Cones | Venus | Average |
---|
SGBM | 12.16 | 22.43 | 14.00 | 10.22 | 14.70 | BM | 14.23 | 21.45 | 10.93 | 11.69 | 14.575 | SGM | 12.78 | 19.46 | 15.61 | 8.92 | 14.19 | Method in Ref.[11] | 6.61 | 6.24 | 4.21 | 5.45 | 5.6275 | Method in Ref.[16] | 1.79 | 11.51 | 8.51 | 0.43 | 5.56 | Method in Ref.[35] | 3.25 | 15.7 | 9.76 | 2.83 | 7.89 | Proposed algorithm | 8.13 | 6.01 | 4.06 | 2.71 | 5.23 |
|
Table 4. False match rate of different algorithms on Middlebury standard dataset