噪声图像合成方法、装置、电子设备及存储介质
By acquiring low-light images and extracting noise data, the problem of unrealistic noise images is solved, and the generated noise images are closer to the real scene, thus improving the training effect of image processing models.
CN117058006BActive Publication Date: 2026-07-17BEIJING ZITIAO NETWORK TECH CO LTD
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING ZITIAO NETWORK TECH CO LTD
- Filing Date
- 2022-05-06
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
In existing technologies, when synthesizing noisy images using noise distribution functions, the noisy images are not realistic, which affects the training effect of image processing models.
Method used
Acquire a low-light image, extract noise data and transfer it to the original image to generate a realistic noisy image.
Benefits of technology
The generated noisy images are more realistic, which improves the training effect of the image processing model.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN117058006B_ABST
Abstract
本公开实施例提供一种噪声图像合成方法、装置、电子设备及存储介质，通过获取暗光图像，暗光图像为低照度场景下拍摄的图像；基于暗光图像进行噪声提取，得到噪声数据；将噪声数据迁移至原始图像中，得到噪声图像。由于本公开实施例中生成噪声图像所使用的噪声数据，是通过获取真实的暗光图像，并基于暗光图像进行提取而得到的，因此更加接近低照度场景拍摄的暗光图像中携带的真实的电子噪声，之后再通过将噪声数据迁移至的原始图像，使生成的噪声图像更加真实，进而提高基于噪声图像进行图像处理模型训练的模型训练效果。
Need to check novelty before this filing date? Find Prior Art