Image reconstruction method and apparatus, terminal device, and storage medium

Patent metadata

Field	Value
Publication number	US-12524837-B2
Application number	US-202118555723-A
Country	US
Kind code	B2
Filing date	Apr 19, 2021
Priority date	Apr 19, 2021
Publication date	Jan 13, 2026
Grant date	Jan 13, 2026

Abstract

The present application relates to image processing technical field, and provides a method for image reconstruction, an apparatus, a terminal device, and a storage medium. The method first extracts an initial feature map of an original image, then calculates an average value of each column pixel in the initial feature map, and constructs a target row vector and duplicates the target row vector in the column direction after convolution processing, to obtain a feature map. In addition, an average value of the element of each row of pixels in the initial feature map is calculated respectively, and a target column vector is constructed. It is duplicated in a row direction to obtain another feature map, and then the two feature maps are fused. Finally, two-dimensional convolution processing is performed on a fused feature map, and a reconstructed image is generated, thereby the long-distance dependencies of the image can be captured.

First claim

What is claimed is:

1 . A method for image reconstruction, comprising:
acquiring a to-be-reconstructed original image; extracting an initial feature map of the original image; respectively calculating an average value of element values for each column of pixels in the initial feature map, and constructing a target row vector in accordance with a calculated average value corresponding to each column of pixels; respectively calculating an average value of element values for each row of pixels in the initial feature map, and constructing a target column vector in accordance with a calculated average value corresponding to each row of pixels; performing one-dimensional convolution processing on the target row vector, and duplicating the target row vector after the one-dimensional convolution processing in direction of column, to obtain a first feature map; performing one-dimensional convolution processing on the target column vector, and duplicating the target column vector after the one-dimensional convolution processing in direction of row, to obtain a second feature map; fusing the first feature map and the second feature map to obtain a third feature map; performing two-dimensional convolution processing on the third feature map to obtain a fourth feature map; generating a reconstructed image corresponding to the initial image in accordance with the fourth feature map.

2 . The method for image reconstruction according to claim 1 , wherein, the step of generating a reconstructed image corresponding to the initial image in accordance with the fourth feature map comprises:
performing a convolution processing with a preset number of convolution kernels on the fourth feature map to obtain a plurality of target feature maps; dividing the plurality of target feature maps into more than two target feature map combinations, wherein, each of the target feature map combinations includes more than two target feature maps; for each target feature map combination, respectively performing two-dimensional convolution processing on each target feature map in the target feature map combination, and then generating a mapping feature map combination corresponding to the target feature map combination by using feature mapping, wherein, the mapping feature map combination includes mapping feature maps obtained after being processed by a preset mapping function for each target feature map in the target feature map combination; generating a reconstructed image corresponding to the initial image in accordance with each target feature map combination and each mapping feature map combination.

3 . The method for image reconstruction according to claim 2 , wherein, the step of generating a reconstructed image corresponding to the initial image in accordance with each target feature map combination and each mapped feature map comprises:
arranging each target feature map combination and each mapped feature map combination in a specified order to obtain a final feature map combination; generating a reconstructed image corresponding to the initial image in accordance with the final feature map combination.

4 . The method for image reconstruction according to claim 3 , wherein, the step of arranging each target feature map combination and each mapped feature map combination in a specified order to obtain a final feature map combination comprises:
arranging each target feature map combination at two ends of the final feature map combination, and arranging each mapped feature map combination between the two ends of the final feature map combination.

5 . The method for image reconstruction according to claim 3 , wherein, the step of generating a reconstructed image corresponding to the initial image in accordance with the final feature map combination comprises:
performing a deconvolution processing on feature maps in the final feature map combination, and then fusing the feature maps after the deconvolution processing to obtain a reconstructed image corresponding to the initial image.

6 . The method for image reconstruction according to claim 2 , wherein, the step of generating a mapping feature map combination corresponding to the target feature map combination by using feature mapping comprises:
using a preset linear function as the mapping function, and performing the feature mapping on each target feature map in the target feature map combination to obtain the mapping feature maps respectively corresponding to each target feature map in the target feature map combination.

7 . The method for image reconstruction according to claim 1 , wherein, the step of fusing the first feature map and the second feature map to obtain a third feature map comprises:
summing elements at corresponding positions of the first feature map and the second feature map to obtain the third feature map.

8 . A terminal device, comprising a memory, a processor, and a computer program stored on the memory and configured to be executed by the processor, wherein, the computer program, when executed by the processor, implements the method for image reconstruction according to claim 1 .

9 . A non-transitory computer readable storage medium, configured to store a computer program, wherein, the computer program, when executed by a processor, implements the method for image reconstruction according to claim 1 .

Assignees

• Shenzhen Inst Of Adv Tech Cas

Inventors

• Wang Shanshan
• Liang Haoyun
• Zheng Hairong
• Liu Xin

Classifications

• G06T2207/20221

  Image fusion; Image merging · CPC title

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T5/50

  using two or more images, e.g. averaging or subtraction · CPC title

• G06T5/60

  using machine learning, e.g. neural networks · CPC title

• G06F18/00

  Pattern recognition · CPC title