A camera imaging quality improvement method based on multi-wavelength weight optical simulation

By constructing a training dataset based on multi-wavelength weighted optical simulation and training a neural network, the problem of image quality degradation caused by the combined effect of multiple wavelength spectral components is solved, thereby improving the image quality and stability of the camera. This method is suitable for tasks such as image deblurring, denoising, and super-resolution.

CN122115242APending Publication Date: 2026-05-29FUDAN UNIVERSITY
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
FUDAN UNIVERSITY
Filing Date
2026-02-03
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing technologies do not adequately consider the characteristics of multi-wavelength spectral composition variations, resulting in insufficient adaptability and stability of camera imaging quality in practical applications, especially when imaging quality deteriorates under the combined effect of visible and near-infrared light.

Method used

By introducing parameters from a real camera optical system, setting multiple different combinations of wavelength weights for optical simulation, generating multi-wavelength weighted simulation images, constructing a training dataset, and training a neural network model, the imaging degradation characteristics under different spectral conditions are learned, thereby improving imaging quality.

Benefits of technology

It improves the adaptability of neural network models to real imaging conditions, enhances the imaging quality and stability of cameras in real-world scenarios, and does not require changes to existing optical hardware structures.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122115242A_ABST
    Figure CN122115242A_ABST
Patent Text Reader

Abstract

The application discloses a camera imaging quality improvement method based on multi-wavelength weight optical simulation. The method comprises the following steps: establishing a model based on the real optical system parameters of a target camera, setting multiple different wavelength weight combinations to represent the relative energy proportion of different wavelengths of light in the imaging process, performing ray tracing simulation on each sub-wavelength light, and generating a simulation image by weighting and superimposing each sub-wavelength imaging result according to the corresponding wavelength weight, thereby generating more diverse simulation data, effectively enhancing the representativeness of the training data for the real imaging scene; and constructing the simulation image and the corresponding original image into a training data set to train a neural network, so that the neural network learns the real imaging degradation law of the target camera under the joint action of different wavelength components and is applied to improve the actual image quality of the target camera. The application can improve the imaging quality and stability of the camera in the real application scene without changing the existing optical hardware structure.
Need to check novelty before this filing date? Find Prior Art