A New Method for Video Inpainting Based on PCA, Neural Network and Ant Colony Algorithms

Digital image inpainting is a process that has been studied over recent years. Image inpainting refers to algorithms that help to fill in deprived or deprived parts of an image or video, remove objects in them, or other desired purposes of the image. Considering the many applications of image inpainting, many researches have been done in this field. However, in the field of video inpainting, research is much less than image inpainting and the results are not representative. Video inpainting requires a large amount of pixels to be covered, and the search space is very large; also, in terms of spatial coordination in a frame, video inpainting must ensure that the time synchronization between the frames is also maintained. In this paper, a PCA-based, neural network and Morpagan algorithm are proposed for video inpainting. In the proposed method, we first consider the reduction of the dimensions of the feature using the PCA dimming algorithm. Then, in the next step, these features are applied with an artificial neural network, as an input to the neural network. In the end, an optimization algorithm has been developed using the Ant s algorithm to reduce the difference between the restored image and the original image. In order to evaluate the proposed method by using the data set of several different videos of similar tasks, the images in the specially created frames were destroyed by Gaussian noise, and then the images were restored using the proposed algorithm, as well as with other methods Waveforms have been compared. The simulation results of the proposed algorithm have better performance than other available methods. The proposed algorithm for the image with 50% noise, the ASSIM is 0.997628 and the ISRE value is 34.68245.

A New Method for Video Inpainting Based on PCA, Neural Network and Ant Colony Algorithms Keywords:

video inpainting , PCA algorithm , neural network , ant colony algorithm , artificial intelligence