Lighting Adaptable Map for Indoor Visual Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for visual localization in indoor environments struggle with varying lighting conditions, particularly in areas with high texture changes or low-textured regions, where lighting changes significantly impact camera tracking and pose estimation, leading to decreased matching quality and sensitivity to illumination changes.
Innovation Solution
A computer-implemented method for building a lighting adaptable map of an indoor scene that adjusts its radiance appearance based on the state of each light source, allowing for explicit exploitation of lighting effects and eliminating the need for reflectance estimation, by obtaining first image information with all light sources on, estimating a map with radiance and reflectance characteristics, detecting and segmenting light sources, and combining their radiance contributions for any given light setting.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If methods target lighting invariance by using brightness normalization and feature descriptors like SIFT or ORB, then robustness to moderate illumination changes is achieved, but matching quality decreases in scenes with high temporal lighting variations and valuable lighting gradients are lost
Solution Approach 1:
Instead of trying to make the system invariant to lighting changes (normalization approach), the patent inverts the approach by making the map adaptive to lighting conditions. The system explicitly models lighting effects and adjusts the map representation to match the current lighting, thereby preserving valuable lighting gradients while maintaining robustness.
Solution Approach 2:
The patent changes the parameters of the map representation from fixed brightness-normalized features to dynamic radiance values that can be adjusted according to lighting conditions. By storing radiance information and light source characteristics, the system can adapt the map's appearance to match current lighting, improving matching quality in temporally varying lighting scenarios.
2Reliability
If multiple feature maps are maintained for different scene appearances or deep learning methods are used, then robustness to lighting changes is achieved given sufficient training data, but device complexity and computational requirements increase
Solution Approach 1:
The patent performs preliminary action by pre-computing and storing radiance contributions of individual light sources and their spatial distributions during map building. This preprocessing allows the system to efficiently adapt to different lighting conditions by simply combining pre-computed light source contributions according to current lighting states, avoiding the need for multiple pre-computed feature maps or complex deep learning models.
3Loss of information
If lighting effects are explicitly exploited for camera tracking, then valuable information in low-textured regions is utilized, but the system becomes sensitive to accurate light source detection and segmentation
Solution Approach 1:
The patent implements feedback by using the detected light setting (state of light sources) to adaptively adjust the map representation. The system detects light sources and their states, then uses this information to weight and combine radiance contributions from different light sources when building the adapted map. This feedback loop allows the system to robustly exploit lighting information even when detection is challenging, as the adaptive mapping compensates for uncertainties.
Data Source
Figure 1~2
Figure 3
AI summary
The invention relates to a computer implemented method for building a lighting adaptable map of an indoor scene, the scene comprising at least one light source and at least one surface, wherein the lighting adaptable map adapts its appearance based on a given light setting, wherein the light setting is defined by the state of each light source, comprising the steps of: • obtaining first image information of the scene, where all light sources are turned on, • estimating a map of the scene based on said first image information, said map comprising radiance information and light reflecting characteristics of the surfaces in the scene, • detecting and segmenting individual light sources in the scene based on the estimated map, • estimating the radiance contribution of each light source to the scene based on the estimated map, and building the lighting adaptable map by storing the estimated radiance contributions and combining them for any given light setting.