An effective inpainting technique for hole filling in DIBR synthesized images

Resumen

This paper proposes an effective image inpainting method for Depth-Image-Based Rendering (DIBR) synthesized images technique to solve disocclusion holes problems in texture images and their respective depth maps. Disoccluded areas, or holes, are classified depending on the size of the hole, being more critical those that appear in the border of the objects, normally between two great depth value differences. Exemplar-based inpainting contains the essential process required to replicate texture and structure for large dissocluded area. The proposed technique focus on preprocessing to improve gradient detection using sobel algorithm followed by a depth-aided exemplar-based technique gives encouraging results.