Contributions to the Automatic Restoration of Images from Scenes in Participating Media
Resumo
This work deals with the problem of image restoration of monocular images acquired in participating media, i.e. media that interfere with light propagation. Specifically, the proposed work focus on the automatic restoration of images acquired in underwater and foggy/hazy scenes. The proposed restoration process requires at least a pair of images as input and produces images in which the medium effects are attenuated and the visibility improved. Differently from previous works, our method does not need additional equipment or information. We proposed a new model-based approach by estimating the depth map and the attenuation coefficient. We performed experimental evaluation in real and simulated environments with significant improvement in the quality of the images.
Referências
He, K., Sun, J., and Tang, X. (2011). Single image haze removal using dark channel prior. IEEE TPAMI, 33(12):2341–2353.
Kim, J., Jang, W., Park, Y., Lee, D., Sim, J., and Kim, C. (2012). Temporally coherent real-time video dehazing. In IEEE ICIP, pages 969–972.
Lowe, D. G. (2004). Distinctive image features from scale-invariant keypoints. IJCV, 60(2):91–110.
Roser, M., Dunbabin, M., and Geiger, A. (2014). Simultaneous underwater visibility assessment, enhancement and improved stereo. In IEEE ICRA, pages 3840–3847.
Zhang, J., Li, L., Zhang, Y., Yang, G., Cao, X., and Sun, J. (2011). Video dehazing with spatial and temporal coherence. The Visual Computer, 27(6-8):749–757.
