Unlock instant, AI-driven research and patent intelligence for your innovation.

A method of optical distortion correction based on deep learning

An optical distortion and deep learning technology, applied in the field of computational photography, can solve problems such as slow solution speed, inability to solve, and inability to accelerate global fast Fourier operations, to reduce calibration requirements and reduce dependencies.

Active Publication Date: 2021-07-30
NANJING UNIV
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Neither of these two types of methods can be solved on the entire non-uniform image, and the global fast Fourier acceleration operation cannot be used, and the solution speed is slow

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • A method of optical distortion correction based on deep learning
  • A method of optical distortion correction based on deep learning
  • A method of optical distortion correction based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0015] The embodiments described below by referring to the figures are exemplary and are intended to explain the present invention and should not be construed as limiting the present invention.

[0016] In this embodiment, an optical distortion correction method based on deep learning, first calibrate the PSF of the lens, and only need to measure a total of about 4 to 7 points at different positions under the data enhancement technology, and the number of points is related to the specific lens type; use the calibrated PSF generates a data set; uses the generated training set to train a specially designed neural network structure; after the training is completed, the trained model can be used to reconstruct the image to be cleared. The specific calculation method steps are as follows:

[0017] Step 1, measure the lens PSF. Use a star hole plate to make a point light source in a dark room, and the hole diameter of the star hole plate is λ 1 , the sensor pixel size is λ 2 , th...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses an optical distortion correction method based on deep learning, comprising the following steps: step 1, calibrate the point spread function PSF of the lens; step 2, use the calibrated point spread function PSF to make a data set through a data generator; step 3. Build a neural network framework: implement three different scale networks through up-and-down sampling convolution. In the residual module, two convolutional layers are stacked and the batch normalization layer is removed. In addition, a discarding layer is added before the convolutional layer; steps 4. Use the generated training set to train the built neural network structure; after the training is completed, you can use the trained model to reconstruct the image to be cleared. The invention utilizes the change law of the point spread function PSF to carry out the data enhancement method, reduces the requirement on the calibration of the point spread function PSF, and also reduces the dependence on the training data set.

Description

technical field [0001] The invention relates to the field of computational photography, in particular to an image non-blind deblurring method. Background technique [0002] Optical distortion is the biggest challenge affecting the imaging quality of imaging systems. Distortion mainly includes spherical aberration, coma, chromatic aberration, and astigmatism. Optical systems generally eliminate distortion by combining multiple lenses with different refractive indices. However, even the most sophisticated optical system cannot completely eliminate these distortions. System designers need to make a trade-off between imaging quality and system complexity. It is difficult to eliminate distortion from the perspective of optical design, and it is expensive and heavy, making it difficult to work in mobile terminals or other environments. [0003] In recent years, with the improvement of computing power, many computing methods have been introduced into image processing. These meth...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): G06T5/00G06T5/50G06T3/40G06T3/60
CPCG06T3/40G06T3/60G06T5/50G06T2207/20021G06T2207/20081G06T2207/20084G06T5/80
Inventor 岳涛徐伟祝曹汛
Owner NANJING UNIV