Improved inpaint-object-remover method with structure tensor
This is the improved implementation method of the algorithm described in the paper "Region Filling and Object Removal by Exemplar-Based Image Inpainting" by A. Criminisi et al.
Requires python 3.5 or newer. You can download it here.
Get inside the project's directory, activate your conda environment(if it`s necessary) and run:
pip install -r requirements.txt
to install the dependencies.
Recommend to use Anaconda to manage your virtual environments
After install this, inside the project's directory and activate your conda environment, run:
pip install -r requirements.txt
to install the other dependencies.
Inside the project's directory run:
python __main__.py [-h] [-ps PATCH_SIZE] [-o OUTPUT] [-p] [-i INPUT_IMAGE] [-m MASK]
You can also pass the -p argument to watch the image being created. For example, use:
python __main__.py -i ../resources/image1.jpg -m ../resources/mask1.jpg
to process image1 inside resources folder using mask1 while printing each result. I provided some images from the paper inside the resources folder.
Use python __main__.py -h to show the available arguments.
It proformance well on a small scale, however, with scale increasing, it gets worst.
Some reslut of repair:
result of image2
result of image3
result of image9
An anvanced version of this programme has been posted on here, it performances better than before.
Ren S, Tang X H, Kang J L. Improved Criminisi algorithm with the texture and edge features[J]. Journal of Image and Graphics, 2012, 17(9): 1085–1091. [任澍, 唐向宏, 康佳伦. 利用纹理和边缘特征的Criminisi改进算法[J]. 中国图象图形学报, 2012, 17(9): 1085–1091. ] [DOI:10.11834/jig.20120906]
He Y T, Tang X H, Zhang Y, Yang R . Improved Criminisi algorithm based on structure tensor[J]. Journal of Image and Graphics, 2018, 23(10): 1492-1507. [何雨亭, 唐向宏, 张越, 杨瑞. 结构张量的改进Criminisi修复[J]. 中国图象图形学报, 2018, 23(10): 1492-1507.][DOI: 10.11834/jig.170650]
Criminisi A, Perez P, Toyama K. Region filling and object removal by exemplar-based image inpainting[J]. IEEE Transactions on Image Processing, 2004, 13(9): 1200–1212. [DOI:10.1109/TIP.2004.833105]