To a Better Digitalization and Visualization of Frontal Face Photographs
André Sobiecki, L.A. Pereira Neves, Carlos Eduardo Thomaz · 2010
Over time human identity photograph may become noisy for digitalization due to its low original resolution, poor quality of the photographic paper and the addition of artifacts on it such as stamps. This article describes a framework based on filtering, segmentation, binarization and inpainting methods to eliminate small stamps artificially added on frontal face images, improving the digitalization and visualization of the original photos. The framework uses information about the relative size of the stamp on the image and its colour to eliminate such artifact and reconstruct the face image.