Noise-Resilient Statistical Framework for Multi-Region Satellite Image Classification Using Run-Length and Gradient-Based Descriptors and a Hybrid Statistical Approach
الملخص
Gaussian noise represents a significant challenge in satellite image processing, as it distorts pixel level relationships and compromises the fidelity of spatial data. This degradation hinders accurate land-cover classification, which is essential for applications such as urban planning, environmental monitoring, and resource management. To address this issue, this study proposes an enhanced hybrid statistical framework designed to classify satellite images into three primary regions (urban, vegetation, and water) even under significant noise interference. The core innovation of this research lies in a dual-stage processing pipeline that integrates texture and structural feature extraction with proactive noise suppression. First, a median filter is applied to mitigate Gaussian noise while preserving critical edge information. Subsequently, two complementary feature descriptors are employed: the Gray-Level Run-Length Matrix (GLRLM), which captures texture patterns through the analysis of consecutive pixel runs, and the Histogram of Oriented Gradients (HOG), which encodes local shape and edge information through gradient orientation distributions. The fusion of these features provides a robust and discriminative representation of land-cover types. The proposed framework was rigorously evaluated using Sentinel-2 imagery of Shanghai, subjected to varying levels of synthetic Gaussian noise (0%, 5%, 10%, and 20%). Performance was assessed using multiple quantitative metrics: Mean Squared Error (MSE), Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), Classification Accuracy (ACC), and Execution Time. Experimental results demonstrate the superiority of the hybrid approach. At 20% noise, the proposed method achieved an exceptional classification accuracy of 98.55%, alongside a PSNR of 122.45 dB and an SSIM of 0.98. In contrast, conventional GLRLM and HOG methods exhibited significant performance degradation under noise, with accuracies fluctuating widely and structural integrity compromised. The hybrid framework also produced a more realistic and balanced distribution of land-cover areas, correctly identifying urban regions that were often misclassified by traditional methods
التنزيلات
المراجع
[1] Aigbokhan, O. J., Essien, N. E., Ogoliegbune, O. M., Afolabi, O. S., & Adamu, I. S. (2022). Assessing Image Classification Accuracy with Principal Component Analysis Algorithm Case Study: Odeda LGA of Ogun State, Southwest Nigeria. Journal of Applied Science and Environmental Management, 26(5), 851–858. https://doi.org/10.4314/jasem.v26i5.11
[2] Al-Rawi, A. G., & Wadood, M. A. (2020). Using multidimensional scaling technique in image dimension reduction for satellite image. Periodicals of Engineering and Natural Sciences (PEN), 8(1), 447–454. https://doi.org/10.21533/pen.v8i1.1171.g527 , http://dx.doi.org/10.21533/pen.v8i1.1171
[3] Ashraf, R., Nisha, R., Shamim, N. F., & Shams, N. S. (2024). Cutting through the noise: A Three-Way Comparison of Median, Adaptive Median, and Non-Local Means Filter for MRI Images. Sir Syed University Research Journal of Engineering & Technology, 14(1), 01–06. https://doi.org/10.33317/ssurj.600
[4] Da Costa, G. B. P., Contato, W. A., Nazaré, T. S., Neto, J. D. E. S. B., & Ponti, M. (n.d.). An empirical study on the effects of different types of noise in image classification tasks. arXiv (Cornell University). https://doi.org/10.48550/arxiv.1609.02781
[5] Azizah, F., Putri, D., Permana, R., Sumarti1, H., & Darma, P. (2025). Classification of CT Scan Images of Stroke Patients and Normal Brain Based on Histogram, GLCM, and GLRLM Texture Features using K-Nearest Neighbor. Journal of Physics and Its Applications, 7(4), 127–134.
[6] D. R., S., Shenoy, P. D., & K. R., V. (2017). Remote Sensing Satellite Image Processing Techniques for Image Classification: A Comprehensive Survey. International Journal of Computer Applications, 161(11), 24–37.
[7] Ding, C., Zhang, M., & Gu, Y. (2021). Study on image quality Control Method based on Gaussian Noise. Journal of Physics Conference Series, 2029(1), 012034. https://doi.org/10.1088/1742-6596/2029/1/012034
[8] Jaber, A., & Abdul Wadood Mohammed, M. (2025). Using factor analysis to Reduction Dimensional of Satellite Images. Journal of Al-Rafidain University College for Sciences (Print ISSN: 1681-6870, Online ISSN: 2790-2293), 56(1), 148-155. https://doi.org/10.55562/jrucs.v56i1.13
[9] Jaber, A., & Abdul Wadood Mohammed, M. (2025). Using Statistical Non-Linear filters for Gaussian Carina Nebula image denoising. Journal of Al-Rafidain University College for Sciences (Print ISSN: 1681-6870 Online ISSN: 2790-2293), 56(1), 431–443. https://doi.org/10.55562/jrucs.v56i1.39
[10] Kairuddin, W. N. H. W., & Mahmud, W. M. H. W. (2017). Texture feature analysis for different resolution level of kidney ultrasound images. IOP Conference Series Materials Science and Engineering, 226, 012136. https://doi.org/10.1088/1757-899x/226/1/012136
[11] Kayathri, K., & Kavitha, K. (2024). CGSX Ensemble: An integrative machine learning and deep learning approach for improved diabetic retinopathy classification. International Journal of Electrical and Electronics Research, 12(2), 669–681. https://doi.org/10.37391/ijeer.120245
[12] Kelly, C., Siddiqui, F. M., Bardak, B., & Woods, R. (2014). Histogram of oriented gradients front end processing: An FPGA based processor approach. Signal Processing Systems (SiPS), 6(2), 1–6. https://doi.org/10.1109/sips.2014.6986093
[13] Kitayama, M., & Kiya, H. (n.d.). Generation of Gradient-Preserving Images allowing HOG Feature Extraction. arXiv (Cornell University). https://doi.org/10.48550/arxiv.2104.01350
[HOG3] Mondal, S. (2018). Hog Feature - A Survey. IOSR Journal of Computer Engineering (IOSR-JCE), 20(4), 1–11. https://doi.org/10.9790/0661-200402011
[14] Lee, K., Choo, C. Y., See, H. Q., Tan, Z. J., & Lee, Y. (2010). Human detection using Histogram of oriented gradients and Human body ratio estimation. Conference: Computer Science and Information Technology (ICCSIT), 2010 3rd IEEE International Conference, 4, 18–22. https://doi.org/10.1109/iccsit.2010.5564984
[15] Mehta, R., & Aggarwal, N. (Eds.). (2014). Comparative Analysis of Median Filter and Adaptive Filter for Impulse Noise - A Review. (2nd ed., Vols. 29–34).
[16] Mohammed Abdul Wadood Mohammed, Assma Ghalib Jaber; A new non-linear filter for image denoising. AIP Conf. Proc. 5 March 2025; 3264 (1): 050061. https://doi.org/10.1063/5.0258639
[17] Novitasari, D. C. R., Lubab, A., Sawiji, A., & Asyhar, A. H. (2019). Application of Feature Extraction for Breast Cancer using One Order Statistic, GLCM, GLRLM, and GLDM. Advances in Science Technology and Engineering Systems Journal, 4(4), 115–120. https://doi.org/10.25046/aj040413
[18] Ochango, V. M., Wambugu, G. M., & Ndia, J. G. (2022). Feature Extraction using Histogram of Oriented Gradients for Image Classification in Maize Leaf Diseases. International Journal of Computer and Information Technology (2279-0764), 11(2). https://doi.org/10.24203/ijcit.v11i2.204
[19] Poobathy, D., & Chezian, R. M. (2014). Edge Detection Operators: Peak signal to noise Ratio Based comparison. International Journal of Image Graphics and Signal Processing, 6(10), 55–61. https://doi.org/10.5815/ijigsp.2014.10.07
[20] Sakhare, & Alase, K. (2020). Noise Models in Digital Image Processing. International Advanced Research Journal in Science, Engineering and Technology, 7(12), 17–21.
[21] Sedeeq, I. (2025). Image forgery detection using Histogram-Oriented Gradients (HOG). Iraqi Journal of Science, 2048–2058. https://doi.org/10.24996/ijs.2025.66.5.22
[22] Singh, K. S. H. R. (2015). International Journal of Computer Engineering and Technology (IJCET) | Computer Engineering and Technology | Journal | © IAEME Publication. http://iaeme.com/Home/issue/IJCET?Volume=7&Issue=6
[23] Soler, J. D., Beuther, H., Rugel, M., Wang, Y., Clark, P. C., Glover, S. C. O., Goldsmith, P. F., Heyer, M., Anderson, L. D., Goodman, A., Henning, T., Kainulainen, J., Klessen, R. S., Longmore, S. N., McClure-Griffiths, N. M., Menten, K. M., Mottram, J. C., Ott, J., Ragan, S. E., . . . Schilke, P. (2019). Histogram of oriented gradients: a technique for the study of molecular cloud formation. Astronomy and Astrophysics, 622, A166. https://doi.org/10.1051/0004-6361/201834300
[24] Vara, G., Rustici, A., Sechi, A., Mosconi, C., Lucidi, V., & Golfieri, R. (2020). Texture analysis on ultrasound: The effect of time gain compensation on histogram metrics and gray-level matrices. Journal of Medical Physics, 45(4), 249. https://doi.org/10.4103/jmp.jmp_82_20
[25] Wang, Z., Bovik, A., Sheikh, H., & Simoncelli, E. (2004). Image Quality Assessment: From Error Visibility to Structural Similarity. IEEE TRANSACTIONS ON IMAGE PROCESSING, 13(4), 1–14.
[26] Zhang, H., Hung, C., Min, G., Guo, J., Liu, M., & Hu, X. (2019). GPU-Accelerated GLRLM Algorithm for feature extraction of MRI. Scientific Reports, 9(1), 10883. https://doi.org/10.1038/s41598-019-46622-w
[27] Zheng, W. (2023). Current Technologies and Applications of Digital Image Processing. Journal of Biomedical and Sustainable Healthcare Applications, 3(1). https://doi.org/10.53759/0088
