A Color Image Completion Method Based on Tensor Block Circular Unrolling

Abstract

The invention discloses a color image completion method based on tensor block cyclic unrolling, and belongs to the technical field of image processing. First, input the image to be completed and initialize the missing pixels with n-nearest neighbors to obtain the target image; then initialize the model parameters, estimate the block cyclic unrolling rank of the target image, and set the weight coefficient. Then, the target image is input into the image completion model in the form of tensor, and the alternating direction multiplier method is used to solve the convex optimization of the model through iteration. The image completion model is a low-rank matrix factorization model based on tensor block cyclic unrolling. . Finally, convert the data format of the iterative tensor so that it is output in the format of the image to be completed. This method increases the connection between image slices when tensor block circular unrolling is performed, thereby reducing the loss of image structure information caused by the unrolling operation to a certain extent; the peak signal-to-noise ratio of the completed image is significantly improved , and the texture and detail information is richer.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a color image completion method based on tensor block cyclic unrolling. Background technique [0002] With the development of the times and the advancement of science and technology, the representation of data is constantly innovating. For image data, the black and white era can be represented and processed by using matrices; however, in the color era, it becomes difficult to process images using matrices, so high-order arrays are used to store images to preserve their special structures. This high-order array Often called a tensor. Tensors are a generalization of the concept of vectors and matrices, which can be regarded as special first-order and second-order tensors. [0003] Visual information is the most effective and most effective information for people to perceive and perceive the world in their daily life. From two-dimensional grayscale images to t...

AI Technical Summary

Problems solved by technology

Tensor decomposition models mainly include CP decomposition, Tucker decomposition, Tensor Train (TT) decomposition, TensorRing (TR) decomposition, and Hierarchical Tensor (HT) decomposition; The original structure of the tensor reduces the connection between tensor elements to a certain extent

As a result, the effect of tensor completion is not satisfactory

Images