Image Processing Using Infrared Guide for Depth Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing methods lack effective means to enhance the quality of depth images without direct depth information, particularly when depth cameras are not used, and the quality and quantity of depth information vary with camera specifications.
Innovation Solution
An image processing method and apparatus that receive an input image and a guide image, extract informative features through supervised or unsupervised learning, and selectively obtain features using attention networks to enhance the input image by aggregating features from both images, ensuring accurate depth information and image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If depth information is obtained from a depth image captured through a depth camera, then spatial distribution of objects can be predicted, but the quality and amount of depth information vary depending on the specification of the depth camera
Solution Approach 1:
The patent introduces a guide image (infrared image) as an intermediary to transfer depth information when direct depth imaging is unavailable or low-quality. The guide image serves as a mediator that provides alternative depth cues through infrared reflection patterns, allowing the system to overcome limitations of specific depth camera specifications.
Solution Approach 2:
The patent transforms the problem from directly measuring depth (which varies by camera spec) to estimating depth from infrared image parameters. By changing the measurement parameter from direct depth values to infrared intensity patterns and surface normal estimates, the system achieves consistent performance across different camera specifications.
2Measurement precision
If informative features are extracted from both input image and guide image to enhance the input image, then image quality and depth information accuracy are improved, but the processing complexity increases
Solution Approach 1:
The patent extracts only the necessary informative features (surface normals, depth estimates, edges) from the guide image rather than processing the entire image. This selective extraction reduces processing complexity while maintaining depth information accuracy by focusing computational resources on the most relevant features.
Solution Approach 2:
The patent segments the feature extraction process into distinct components: infrared image processing, surface normal estimation, depth map generation, and feature aggregation. This segmentation allows each component to be optimized independently, reducing overall processing complexity while improving depth information accuracy.
Data Source
AI summary
An image processing method includes receiving an input image and a guide image corresponding to the input image, extracting informative features from the input image and the guide image to enhance the input image, selectively obtaining a first feature for the input image from among the informative features, and processing the input image based on the first feature.