نوع مقاله : علمی - پژوهشی
نویسندگان
1 استاد گروه سنجش از دور ، دانشکده مهندسی نقشهبرداری دانشگاه صنعتی خواجه نصیرالدین طوسی
2 کارشناس ارشد مهندسی برق، مخابرات مؤسسۀ آموزش عالی خاوران
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
Detectors noises in satellite images are seen as either vertical or horizontal stripes. The directions of these stripes depend on the imaging technique (Pushbroom or Wiskbroom). The main reasons in appearance of stripe noises in TM images are; lack of matching between detectors, unsuitable calibration and detector degradation in time. Due to the Wiskbroom technique in TM sensor, the stripes appear horizontally. Among these, the stripe noises in band4 are more profound in images acquired from dark surfaces such as sea surface. This kind of noise may produce sever errors in atmospheric correction based on dark surfaces. In this work, to correct the stripe noise, Mean Method (MM), Modified Spatial Momentum Matching (MSMM), and image filtering in frequency and spatial domain (IFFD & IFSD) are introduced. To evaluate the results, some statistical parameters such as averaging, standard deviation, histograms and Fourier spectrums before and after corrections are deployed. Reduction in standard deviation after denoising demonstrates enhancement in the image. To compare these methods with other known methods, parameters such as MSE, RMSE and PSNR along with simulated images for periodical striped noise are used. Among these, the maximum PSNR and naturally the minimum MSE belongs to MM and MSMM methods and consequently these methods perform better accuracies compared to IFFD and IFSD.
کلیدواژهها [English]