• Laser & Optoelectronics Progress
  • Vol. 57, Issue 10, 101020 (2020)
Qingqing Ma and Caifang Wang*
Author Affiliations
  • College of Arts and Sciences, Shanghai Maritime University, Shanghai 201306, China
  • show less
    DOI: 10.3788/LOP57.101020 Cite this Article Set citation alerts
    Qingqing Ma, Caifang Wang. Non-Local Stereo Matching Algorithm Based on Color and Edge Information[J]. Laser & Optoelectronics Progress, 2020, 57(10): 101020 Copy Citation Text show less
    Flow chart of proposed algorithm
    Fig. 1. Flow chart of proposed algorithm
    Schematic of cost aggregation in minimum spanning tree structure. (a) From leaf nodes to root nodes; (b) from root nodes to leaf nodes
    Fig. 2. Schematic of cost aggregation in minimum spanning tree structure. (a) From leaf nodes to root nodes; (b) from root nodes to leaf nodes
    Experimental results of proposed algorithm and other algorithms. (a) Original images; (b) true disparity images; (c) disparity images of traditional algorithm; (d) disparity images of proposed algorithm; (e) mismatch of proposed algorithm
    Fig. 3. Experimental results of proposed algorithm and other algorithms. (a) Original images; (b) true disparity images; (c) disparity images of traditional algorithm; (d) disparity images of proposed algorithm; (e) mismatch of proposed algorithm
    Robustness test results for different parameters. (a) λ; (b) α
    Fig. 4. Robustness test results for different parameters. (a) λ; (b) α
    Parameterβτ1τ2αλσN
    Value0.117.002.000.160.500.104.00
    Table 1. Experimental parameters
    AlgorithmTsukubaVenusTeddyConesAverage
    NonoccAllDiscNonoccAllDiscNonoccAllDiscNonoccAllDisc
    Proposed1.541.837.690.400.662.924.619.7812.242.898.388.285.10
    NLCA[10]2.943.469.380.330.583.356.409.1315.433.468.049.776.02
    VSW[17]1.621.886.980.470.813.408.6713.3018.003.378.828.126.29
    ASW[6]1.381.856.900.711.196.137.8813.3018.603.979.798.266.67
    VC[18]1.992.656.770.620.963.209.7515.1018.206.2812.7012.907.60
    Table 2. Comparison of mismatched pixels with threshold is 1unit: %
    AlgorithmTsukubaVenusTeddyConesAverage
    NonoccAllDiscNonoccAllDiscNonoccAllDiscNonoccAllDisc
    Proposed13.4613.7615.2811.0811.5413.9812.6718.3822.5011.7917.1417.7214.94
    NLCA[10]11.1811.7515.399.399.2917.9315.6118.2230.7311.7716.1222.2915.84
    VSW[17]19.2019.5018.508.178.6513.2017.4023.2031.4013.1018.3020.4017.60
    ASW[6]18.1018.8018.607.778.4015.8017.6023.9034.0014.0019.7020.6016.03
    VC[18]24.5025.1021.509.039.5913.8018.8025.1031.4016.1022.1022.4019.90
    Table 3. Comparison of mismatched pixels with threshold is 0.5unit: %
    AlgorithmTsukubaVenusTeddyConesAverage
    Proposed3.691.338.886.545.10
    NLCA5.261.4210.327.096.02
    Error improvement29.856.3413.957.7615.28
    Table 4. Error reduction comparisonunit: %
    Qingqing Ma, Caifang Wang. Non-Local Stereo Matching Algorithm Based on Color and Edge Information[J]. Laser & Optoelectronics Progress, 2020, 57(10): 101020
    Download Citation