PATCH BASED IMAGE INPAINTING USING TEXTON HISTOGRAMS AS CONTEXTUAL DESCRIPTORS
Author’s Name : Abhijeet Chaudhari | Rupesh Dheple | Evans Kurapaty | Aniket Mulmule
Volume 03 Issue 03 Year 2016 ISSN No: 2349-3828 Page no: 1-3
Image In painting is an image processing task of filling in the missing region in an image in a visually plausible way. The current methods of image inpainting consider only a segment of the image to obtain relevant data and fill the target patch. The proposed method uses patch based image inpainting, also known as textural inpainting. Textural inpainting methods fill in the missing region patch-by-patch by searching for well-matching replacement patches (i.e., candidate patches) in the undamaged part of the image and copying them to corresponding locations. Texton histograms, which give similarities between any two patches are referred to for selection of candidate patches. In the process, a huge number of candidate patches will be selected. The other drawback of existing inference methods is their inefficiency in selecting the right patch when the number of candidate patches is huge. In order to overcome this drawback, a novel optimization approach suitable for the inpainting problem with large number of patches has been proposed. This optimized approach is based on Markov Random Field Modelling.
Inpainting, patch-based, texture features, context-aware.
- Kaiming He, Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition.
- J Puzicha, Histogram Clustering for Unsupervised Image Segmentation.
- P Arbelaez, Contour Detection and Hierarchical Image Segmentation.
- N Komodakis, Image Completion Using Efficient Belief Propagation Via Priority Scheduling and Dynamic Pruning.
- A Antonio, Context Aware Patch Based Image Inpainting Using Markov Random Field.