Image Processing Exposure Sequencing for High-Contrast Visual SLAM
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Visual SLAM systems face challenges in self-localization accuracy due to geometric derivation from camera images, particularly in environments with high contrast, leading to blown-out highlights and blocked-up shadows, which hinder feature point extraction and alignment.
Innovation Solution
An image processing device with an exposure control unit that alternates between positive and negative exposure corrections to stabilize feature point detection, incorporating a Δ pose estimating unit for integrating pose differences from corrected image frames, and an integration unit for filtering pose estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If proper exposure is used for capturing camera images, then image quality is maintained, but in high-contrast environments this leads to blown-out highlights and blocked-up shadows that hinder feature point extraction
Solution Approach 1:
The patent segments the image capture process into multiple exposures (proper exposure, positive correction exposure, and negative correction exposure) to capture different luminance ranges. This segmentation allows feature points to be extracted from images where the luminance is within an appropriate range, avoiding the problems of blown-out highlights and blocked-up shadows that occur with single proper exposure in high-contrast environments.
Solution Approach 2:
The patent changes the exposure parameter by performing positive and negative corrections to the proper exposure. This creates a sequence of image frames with varying exposure levels, allowing the system to select frames where feature points can be successfully extracted based on their luminance characteristics, thereby resolving the contradiction between maintaining proper exposure and ensuring feature point extractability.
2Reliability
If multiple image frames with different exposures are captured, then feature point detection stability is improved, but device complexity increases due to additional processing requirements
Solution Approach 1:
The patent performs preliminary exposure corrections by capturing multiple image frames with predetermined exposure adjustments (positive and negative corrections) before feature point extraction. This preliminary action ensures that at least some of the captured frames will have appropriate luminance for reliable feature point detection, reducing the need for complex real-time processing and selection algorithms.
Solution Approach 2:
The patent creates copies of the image data at different exposure levels (positive correction copies and negative correction copies) from the original captured frames. This copying approach allows the system to evaluate multiple exposure variants and select the most suitable ones for feature point extraction, improving detection stability while maintaining manageable processing complexity through systematic duplication.
Data Source
AI summary
An image processing device (1) includes an exposure control unit (102), a Δ pose estimating unit (109), and an integration unit (111). The exposure control unit (102) controls exposure by sequentially performing positive correction of increasing exposure from proper exposure or negative correction of decreasing exposure from the proper exposure as the exposure at a time of acquiring each of a plurality of image frames in time series in a predetermined execution order. The Δ pose estimating unit (109) estimates first position pose of the image processing device based on matching between image frames subjected to the positive correction and second position pose of the image processing device based on matching between image frames subjected to the negative correction. The integration unit (111) integrates the first position pose and the second position pose.