Depth estimation-based HDR image reconstruction based on underexposure LDR image

A technology of depth estimation and image reconstruction, applied in image enhancement, image analysis, image data processing and other directions, can solve problems such as HDR reconstruction

Pending Publication Date: 2022-01-25
TIANJIN UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

First, HDR reconstruction is combined with depth estimation to solve the HDR reconstruction problem

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
  • Depth estimation-based HDR image reconstruction based on underexposure LDR image

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0006] The present invention combines the knowledge of HDR reconstruction and depth estimation to design the model. The core idea of ​​the present invention is to integrate depth estimation into the HDR image reconstruction model, and develop two connected deep neural networks to solve each task. Specific steps are as follows:

[0007] The first step is the HDR reconstruction network.

[0008] The method of directly inferring 32-bit HDR images from 8-bit LDR images is difficult to train with unsupervised learning. The HDR reconstruction network does not directly map LDR to HDR images, but uses an indirect method to generate LDR images with different exposures, and then Merge it into the HDR image. The specific method is to give the input LDR image, infer the LDR image with higher exposure than the input image from the given underexposed image, then output the overexposed image, and finally generate the final HDR image from these surrounded LDR images .

[0009] The second ...

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

A digital camera can only capture the brightness of a real scene in a limited range. Due to quantization and saturation of a camera sensor, texture details in high dynamic range (HDR) data are lost, and great difficulty is brought to subsequent image-based application (depth estimation and the like). As an index of the HDR image reconstruction effect, the depth estimation error caused by the incorrect hue reflected in the LDR image can support the improvement of the learning efficiency of the HDR image generation network. Therefore, an end-to-end framework composed of two connected CNNs is provided, and the end-to-end framework is used for solving the key problem in HDR image reconstruction. A large number of quantitative and qualitative experiments are carried out on benchmark and recently published challenging data sets, and it is proved that compared with the latest single-image HDR reconstruction and depth estimation algorithm, the proposed method achieves extraordinary performance.

Description

technical field [0001] The invention relates to a reconstruction technology of an HDR image, and depth estimation is introduced into an image reconstruction model, and the whole network is superior to other HDR image reconstruction methods in an unsupervised learning manner. Background technique [0002] Digital cameras can only capture the brightness of real scenes to a limited extent. Due to the quantization and saturation of the camera sensor, the loss of texture details present in high dynamic range (HDR) data brings great difficulties to subsequent image-based applications (depth estimation, etc.). An image with proper brightness and hue is the key to estimating the depth of the scene. Unlike the existing learning-based methods, the method proposed in this paper introduces depth estimation into the learning model of HDR image reconstruction and promotes each other. As an indicator of the HDR image reconstruction effect, the depth estimation error caused by the incorrec...

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 Applications(China)
IPC IPC(8): G06T5/00G06T7/50G06N3/04G06N3/08
CPCG06T5/009G06T7/50G06N3/088G06T2207/20081G06T2207/20084G06T2207/20208G06N3/045
Inventor 张涛梁杰王昊
Owner TIANJIN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap