Inpainting Technique application to Rebuild Partially Detected Dark Slope Streaks/ Aplicação de técnica de inpanting para reconstrução de dark slope streks parcialemtne detectados

• Autor(a) principal: Figueira, Ana Luisa Chaves
• Data de Publicação: 2020
• Outros Autores: Cruz, Breno Strogueia Maia da, Silva, Erivaldo Antonio da, Cardim, Guilherme Pina, Statella, Thiago, Pina, Pedro
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Brazilian Applied Science Review
• Texto Completo: https://ojs.brazilianjournals.com.br/ojs/index.php/BASR/article/view/13217
• topic: Remote Sensing; Mars; Dark Slope Streaks; Cartography; Image Processing; Inpainting.

This paper approaches the implementation of an algorithm that automatically detects features of interest on the Martian surface known as Dark Slope Streaks (DSS), and the application of an inpainting technique to improve the results obtained by the detection. The greatest difficulty for the development of this algorithm is the false positives caused by shadows, depressions or fading of DSS, after a period. To overcome these problems, techniques such as image segmentation and custom filtering routines were used. Sometimes, the extraction algorithm obtains partially detected features, culminating in a loss of quality. However, in order to remove occlusions and restore lost features, the resulting image goes through an inpainting process proposed by Galerne et al (2017). To understand if the inpainting technique improves the quality of the extraction process, we created a reference image and compared with both the image obtained from the extraction algorithm and the resulting image of the inpainting algorithm. This process was repeated for four interest areas. Comparing the mean of pre-reconstruction SSIM and the mean of post-reconstruction SSIM, the inpainting algorithm shows an improvement of 6.13% in quality. Therefore, both algorithms may contribute greatly to the increasing quality of extraction of DSS in the area of cartography.