Edge missing image inpainting with compression–decompression network in low similarity images

Image inpainting technology can patch images with missing pixels. Existing methods propose convolutional neural networks to repair corrupted images. The network extracts effective pixels around the missing pixels and uses the encoding–decoding structure to extract valuable information to repair the vacancy. However, if the missing part is too large to provide useful information, the result will be fuzzy, color mixing, and object confusion. In order to patch the large hole image, we propose a new algorithm, the compression–decompression network, based on the research of existing methods. The compression network takes responsibility for inpainting and generating a down-sample image. The decompression network takes responsibility for extending the down-sample image into the original resolution. We use the residual network to construct the compression network and propose a similar pixel selection algorithm to expand the image, which is better than using the super-resolution network. We evaluate our model over Places2 and CelebA data set and use the similarity ratio as the metric. The result shows that our model has better performance when the inpainting task has many conflicts.