• Laser & Optoelectronics Progress
  • Vol. 56, Issue 21, 211504 (2019)
Deqiang Cheng1、*, Huandong Zhuang1、**, Wenjie Yu1, Chunmeng Bai1, and Xiaoshun Wen2
Author Affiliations
  • 1School of Information and Control Engineering, China University of Mining and Technology, Xuzhou, Jiangsu 221116, China
  • 2Wanbei Coal and Electricity Group Co., Ltd., Suzhou, Anhui 234000, China
  • show less
    DOI: 10.3788/LOP56.211504 Cite this Article Set citation alerts
    Deqiang Cheng, Huandong Zhuang, Wenjie Yu, Chunmeng Bai, Xiaoshun Wen. Cross-Scale Local Stereo Matching Based on Edge Weighting[J]. Laser & Optoelectronics Progress, 2019, 56(21): 211504 Copy Citation Text show less
    Diagram of edge transform
    Fig. 1. Diagram of edge transform
    Teddy's local disparity maps. (a) Disparity map before weighting; (b) disparity map after weighting
    Fig. 2. Teddy's local disparity maps. (a) Disparity map before weighting; (b) disparity map after weighting
    Disparity maps for different algorithms. (a) Ground truth maps; (b) CT-GF; (c) CGA-GF; (d) GA-MST; (e) GA-ST; (f) proposed algorithm
    Fig. 3. Disparity maps for different algorithms. (a) Ground truth maps; (b) CT-GF; (c) CGA-GF; (d) GA-MST; (e) GA-ST; (f) proposed algorithm
    Experimental results of proposed algorithm. (a) Left images; (b) ground truth maps; (c) disparity maps of proposed algorithm (without disparity refinement); (d) disparity maps of proposed algorithm (with disparity refinement); (e) mismatched maps of proposed algorithm (with disparity refinement)
    Fig. 4. Experimental results of proposed algorithm. (a) Left images; (b) ground truth maps; (c) disparity maps of proposed algorithm (without disparity refinement); (d) disparity maps of proposed algorithm (with disparity refinement); (e) mismatched maps of proposed algorithm (with disparity refinement)
    Experimental results of different parameter settings. (a) k1; (b) k2; (c) λn; (d) λs
    Fig. 5. Experimental results of different parameter settings. (a) k1; (b) k2; (c) λn; (d) λs
    Parameterτcτgαk1k2λnλsTSλr
    Value0.027450.007840.110.90.70.20.12040.759
    Table 1. Experimental parameters
    AlgorithmTsukubaVenusTeddyConesAverage
    GF2.621.728.253.584.04
    S+GF2.301.096.993.233.40
    Proposedalgorithm2.120.916.593.063.17
    Table 2. Error matching rates before and after the improvement%
    AlgorithmTsukubaVenusTeddyConesAverage
    S+GF0.811.565.085.433.22
    S+BF37.6070.84232.09231.78143.08
    Proposedalgorithm0.931.725.445.743.46
    Table 3. Comparison of run time of different algorithmss
    Stereo pairsCT-GFCGA-GFGA-MSTGA-STProposed algorithm
    Tsukuba3.542.921.762.042.12
    Venus1.992.501.241.410.91
    Teddy8.568.235.736.226.59
    Cones4.873.754.424.763.06
    Aloe7.045.804.884.835.57
    Art12.159.7910.6910.388.56
    Baby13.313.318.214.493.56
    Baby24.033.7913.5415.122.76
    Baby34.975.285.593.963.85
    Books9.969.1310.6610.048.51
    Bowling16.187.4619.5618.888.75
    Bowling28.476.9010.1110.535.27
    Cloth11.961.160.630.691.29
    Cloth24.333.564.354.353.53
    Cloth32.802.052.902.962.28
    Cloth42.381.951.881.841.68
    Dolls7.275.715.895.494.80
    Flowerpots9.6712.9216.7912.508.54
    Lampshade110.6411.639.819.137.10
    Lampshade210.5215.9712.089.8912.46
    Laundry18.4518.5011.9211.9012.03
    Midd133.4537.6124.4322.0132.90
    Midd232.9635.5320.5718.9026.61
    Moebius10.9210.757.577.308.11
    Monopoly18.9923.4321.0320.7521.68
    Plastic22.2329.3945.0237.3925.74
    Reindeer8.159.369.797.836.05
    Rocks14.033.943.353.063.00
    Rocks22.742.202.281.991.53
    Wood14.934.9910.185.463.66
    Wood22.952.963.174.841.78
    Average matching error rate9.189.7610.009.067.88
    Average rank3.613.423.392.681.93
    Table 4. Average matching error rates of different algorithms on non-occlusion regions%
    AlgorithmTsukubaVenusTeddyConesAverage
    Non-occAllDiscNon-occAllDiscNon-occAllDiscNon-occAllDisc
    GC+occ[5]1.192.016.241.642.196.7511.217.419.85.3612.413.08.26
    Adapt Weight[10]1.381.856.900.711.196.137.8813.318.63.979.798.266.67
    TiwceGF[17]1.822.997.350.341.533.456.9314.7416.083.212.2911.456.84
    ASSW[21]1.812.177.850.320.513.737.0212.517.43.218.408.996.16
    iFBS[22]1.782.107.570.310.502.177.9412.817.13.078.738.466.05
    SDDS[23]3.313.6210.40.390.762.857.6513.019.43.9910.010.87.19
    VSW[24]1.621.886.980.470.813.408.6713.318.03.378.858.126.29
    Proposed algorithm2.022.298.660.260.482.886.1111.415.772.948.808.075.81
    Table 5. Percentage of mismatching pixels in different regions for different algorithms%
    Deqiang Cheng, Huandong Zhuang, Wenjie Yu, Chunmeng Bai, Xiaoshun Wen. Cross-Scale Local Stereo Matching Based on Edge Weighting[J]. Laser & Optoelectronics Progress, 2019, 56(21): 211504
    Download Citation