سنجش كارايي ويژگي‌هاي بافتي GLCM در افزايش دقت طبقه‌‌بندي تصاوير حاصل از ادغام تصاوير تك‌باند و ابرطيفي مناطق مسكوني و صنعتي جنوب شهر تهران

GLCM Texture Features Efficiency Assessment of Pansharpened Hyperspectral Image Classification for Residential and Industrial Regions in Southern Tehran

اكثر الگوریتم‌‌های طبقه‌‌بندی داده‌های سنجش از دور بر اساس ویژگی‌‌ها و اطلاعات طیفی پیكسل‌‌ها عمل می‌كنند. این مسئله باعث نادیده گرفتن اطلاعات مكانی مفید قابل استخراج از این تصاویر، مانند؛ بافت تصاویر می‌شود. استفاده هم‌‌زمان از بافت و اطلاعات طیفی مبحثی است كه به آن كمتر پرداخته شده است. در این پژوهش تاثیر استفاده از بافت تصویر تك‌‌باند سنجنده  ALI  بر دقت طبقه‌‌بندی تصاویر ابرطیفی سنجنده هایپریون در محیط‌‌های شهری بررسی شده است. اطلاعات بافت تصویر تك‌باند با استفاده از ماتریس رخداد همزمان (GLCM) استخراج شده است. طبقه‌‌بندی نیز با به‌كارگیری روش ماشین بردار پشتیبان (SVM) و در سه مرحله انجام پذیرفت: طبقه‌‌بندی اطلاعات طیفی تصویر ابرطیفی، طبقه‌بندی تصاویر ادغام شده به دست آمده از روش تبدیل رنگ نرمالیزه (CNT)، و نهایتا طبقه‌‌بندی با استفاده هم‌‌زمان از بافت تصویر تك‌باند و تصاویر ابرطیفی ادغام شده. تاثیر نوع ویژگی بافت استخراجی از ماتریس رخداد همزمان و همچنین اندازه پنجره استخراج بافت در پوشش‌های مختلف بررسی شد. نتایج پیاده‌سازی‌ها نشان داد كه استفاده از ویژگی‌های بافتی در كنار ویژگی‌های طیفی تصاویر حاصل از ادغام، می‌تواند دقت طبقه‌بندی مناطق شهری، مانند؛ بافت مسكونی و مناطق صنعتی را  به طور كلی، حدود 5 درصد بهبود ببخشد. افزایش دقت در برخی از كلاس‌‌ها تا حدود 15 درصد بوده است.

Most of common classification algorithms in remote sensing are based on spectral characteristics of pixels. These approaches ignore the spatial information of data, such as texture, in classification process. Simultaneous usage of texture and spectral information is a new trend in remote sensing image classification which has been considered in this study. We have evaluated the efficiency of gray-level co-occurrence matrix texture features (GLCM) extracted from panchromatic (PAN) image of ALI detector in improving the classification accuracy of Hyperion hyperspectral (HS) data in urban regions of Tehran. Classification is performed using a support vector machine (SVM) classifier with a Gaussian kernel. In our experiments, we have considered three different cases: a) classifying original Hyperion data, b) classifying Hyperion data pansharpened by color normalized transform (CNT), and finally, c) simultaneous use of GLCM texture features of panchromatic data and the spectral features of pansharpened data in classification process. In case b and c, for pansharpening HS data we have performed the following steps: Registering HS data with PAN data using nineteen ground control points, polynomial warping of second order, and Nearest neighbor interpolation. Selecting a subset of HS bands which spectrally overlap with PAN image. Fusing the spatial information of PAN image into the HS subset bands, obtained in step 2, using CNT method. Moreover, as GLCM features, we have extracted 8 texture features from GLCM matrices: mean, variance, homogeneity, contrast, dissimilarity, entropy, angular second moment, and correlation. In order to assess the influence of the size of GLCM extraction window on the quality of texture features, we have considered various window sizes: 3×3, 5×5, 7×7, and 9×9. At the first phase of our experiments, we compared classification results obtained using the original HS data with the results obtained from the pansharpened HS subset – see Table 1. The results showed an increase of about 15% in the average classification accuracy when using the pansharpened data. At the second phase, we combined each of the texture features individually with the pansharpened HS subset. The results are given in Table 2. As the table suggests, regardless of the type of texture feature and the size of the GLCM extraction window, the combinations improve the overall classification accuracy (OA) of data. However, texture features show better quality when extracted from GLCM matrices obtained using 9×9 neighborhood windows. In addition, we observe that regardless of the size of GLCM window, dissimilarity feature delivers the best results. To summarize, by using the pansharpened HS subset instead of the original HS data, we achieved about 15% gain in the classification accuracy. Moreover, combining dissimilarity texture features –extracted from GLCM matrices obtained using 9×9 neighborhood windows– with the pansharpened HS subset improved classification results. In our experiments, we achieved about 5% increase in OA compared to that of using pansharpened HS subset alone. Table 1. Comparison of classification accuracies obtained using original HS data with those obtained from pansharpened HS subset. Class number Original HS data Pansharpened HS subset Increase in accuracy 1 91.4 99.7 8.3 2 96.3 97.0 0.7 3 99.1 99.6 0.6 4 61.9 70.1 8.2 5 39.4 40.6 1.2 6 69.2 45.8 -23.4 7 1.9 0.6 -1.3 8 98.1 98.1 0 9 67.3 71.3 4.0 10 21.4 76.9 55.5 11 0.0 87.2 87.2 12 99.4 99.2 -0.2 13 99.1 95.4 -3.7 14 3.7 73.2 69.5 Average 60.6 75.3 14.7 Table 2. Comparison of overall classification accuracies (OA) of “pansharpened HS subset” with those of “pansharpened HS subset + GLCM texture feature*”. *GLCM texture feature (combined with pansharpened HS subset) GLCM extraction window size 9×9 7×7 5×5 3×3 Mean 90.1 88.5 87.3 86.6 Variance 87.6 86.6 86.6 86.3 Homogeneity 89.7 88.7 88.7 86.9 Contrast 86.8 86.6 86.4 86.4 Dissimilarity 90.5 90.2 90.0 88.3 Entropy 90.5 89.0 88.9 87.1 Angular second moment 90.1 88.5 88.6 87.3 Correlation 87.3 87.4 88.0 87.3 Pansharpened HS subset (no GLCM features) 86.1