An image completion method based on adaptive rank estimation Riemannian manifold optimization

Abstract

The invention discloses an image completion method based on adaptive rank estimation Riemannian manifold optimization. The method is characterized by on the basis of an L1 norm regularization matrix completion algorithm (L1MC), introducing an adaptive rank estimation penalty term and a Riemannian manifold of a fixed rank; using a penalty item R; correcting an iterative optimization function f (R)and a sequence of the matrix determined by an empirical formula continuously to estimate the sequence Rq of the missing matrix, and utilizing Rq to construct a Riemannian manifold of a fixed rank formatrix completion. In the process, a soft threshold algorithm is avoided, the operation rate is increased, and the operation space is saved. Riemannian manifold optimization is applied, the rank is limited in a range of R < = Rq to find an estimation value closest to an original matrix, and the introduction of a shrinkage operator for reducing the rank in each iteration is avoided, so that the calculation efficiency and the image recovery rate are greatly improved. Besides, the algorithm uses the convolutional neural network to preprocess and construct the low-rank matrix, so that the missingvalue is effectively estimated in a short time, the image recovery process is accelerated, and the image recovery accuracy is improved.

Description

【Technical field】 [0001] The invention belongs to the technical field of image completion, and in particular relates to an image completion method based on adaptive rank estimation Riemannian manifold optimization. 【Background technique】 [0002] Image completion technology has increasingly become a research trend in computer vision and image processing. This technology aims to restore and complement missing pixels from the surrounding structure and texture information of missing pixels, and then obtain a restored image. In practical applications, image loss, damage and noise pollution are generally unavoidable. Taking the wireless transmission of images as an example, even if the image can avoid distortion during transmission, the complete image is often artificially sampled and compressed in order to increase storage space utilization and extend the service life of the image sensor. Image completion technology has a wide range of applications, such as historical image res...

AI Technical Summary

Problems solved by technology

Someone proposed an algorithm that uses Riemannian optimization for low-rank matrix completion. This algorithm uses a Riemannian manifold with a fixed rank to complete the matrix. However, in this algorithm, the fixed rank of the Riemannian manifold is generally difficult to obtain directly. Therefore, the algorithm needs a large number of experiments to solve the appropriate rank number, resulting in low computational efficiency of the algorithm

Someone proposed an image completion algorithm based on OptSpace, and someone proposed an image completion algorithm based on IALM (Inexact Augmented Lagrange Multipliers). Although these two algorithms fully consider the problem of low-rank estimation, the resulting image completion Poor visuals

[0004] Due to the randomness of the distribution of missing pixels in the image, existing algorithms cannot provide a local scene structure that prevents blurred boundaries, and are not fully suitable for pre-visualization of path tracing

When the missing rate of the missing image is high, the restored image reconstructed by the existing algorithm has the disadvantages of low accuracy and low recovery efficiency.

Images