Performance of denoising algorithms in the improvement of lithological discrimination

• Published in: Modeling earth systems and environment 2022-11, Vol.8 (4), p.5381-5388
• Main Authors: El Alaoui El Fels, Abdelhafid, Ben-loghfyry, Anouar, El Ghorfi, Mustapha
• Format: Article
• Language: English
• Subjects: Chemistry and Earth Sciences; Computer Science; Earth and Environmental Science; Earth Sciences; Earth System Sciences; Ecosystems; Environment; Math. Appl. in Environmental Science; Mathematical Applications in the Physical Sciences; Original Article; Physics; Statistics for Engineering
• ISSN: 2363-6203; 2363-6211
• DOI: 10.1007/s40808-022-01401-x

Satellite multispectral systems are fundamental and crucial data sources for the application of spatial classification methods, such as PCA. This statistical approach focuses first on the constitution of new bands that maximize the information in decreasing order. It also focuses on the assignment of the bands of high inertia to the main RGB color to generate a diversified color composite that can be capable of discriminating the lithology. The objective of this study is to extract information from noisy bands and more precisely from degraded bands with low inertia using the image denoising method by total variation regularization (TVR). The study is based on a comparative approach between the two algorithms: finite difference (with fixed-point iterations) and the Primal–Dual. The experimental results obtained with multispectral datasets confirm the validity of this TVR denoising technique and the superiority of the Primal–Dual algorithm. The remarkable obtained results show clear colored compounds and prove that the low inertia bands produced by PCA are capable of providing additional information.