高光谱遥感影像纹理特征提取的对比分析

邵文静,孙伟伟,杨刚

Comparison of Texture Feature Extraction Methods for Hyperspectral Imagery Classification

Wenjing Shao,Weiwei Sun,Gang Yang

表1 印第安纳数据的分类精度 (%)

Table 1 Classification accuracy of Indian Pines data

地物类别 纹理特征提取方法/% riLBP SLIC EMP DMP AP 3D-Gabor JBF GF SVM 苜蓿 95.12 84.88 97.56 98.54 94.02 29.02 62.44 29.76 28.78 玉米未耕地 98.39 91.46 97.32 96.76 95.98 62.37 82.11 79.81 55.83 玉米疏耕地 98.84 92.64 98.53 98.26 97.07 60.13 64.31 63.27 47.20 玉米 98.75 82.82 96.90 99.06 94.38 49.30 84.84 63.99 31.46 牧草地 98.01 93.68 98.21 97.20 96.77 87.38 93.56 94.16 83.52 草树混合地 99.34 97.63 99.88 99.48 99.08 98.22 100.00 100.00 90.91 修剪的牧草 93.33 95.27 99.20 97.60 96.53 74.00 94.00 96.00 56.80 干牧草 99.84 98.51 100.00 100.00 99.59 97.56 100.00 100.00 93.93 燕麦 98.15 77.22 100.00 100.00 93.84 26.67 8.33 1.11 26.67 大豆未耕地 98.74 92.94 97.92 97.53 96.78 62.82 86.13 76.46 58.49 大豆疏耕地 99.37 95.45 99.61 99.33 98.44 78.37 95.27 95.50 78.79 大豆已耕地 98.00 84.10 98.13 97.45 94.42 47.75 72.90 68.16 40.37 小麦 100.00 93.48 99.13 99.46 98.02 96.52 100.00 99.89 93.04 树木 99.85 98.32 100.00 99.98 99.54 95.40 99.89 99.93 94.39 林间小道 99.14 91.87 100.00 99.83 97.71 55.88 60.72 68.27 32.02 钢铁塔 94.84 89.17 96.43 100.00 95.11 86.55 93.81 99.64 80.12 OA 98.98 94.02 98.88 98.64 94.23 74.84 88.03 86.23 69.66 AA 98.11 91.54 98.68 98.78 96.70 69.25 81.14 77.25 62.02 KC 98.83 93.17 98.72 98.45 93.42 71.12 69.70 65.26 65.08