Distance Information Processing Apparatus Depth Range Determination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing imaging apparatuses face challenges in accurately determining the depth range of objects due to incorrect correlation evaluations, particularly when contrast changes little or noise is high, and when large collation regions include multiple objects at different distances, leading to incorrect object distance calculations.
Innovation Solution
A distance information processing apparatus that calculates object distances for multiple pixel positions using first and second image signals, includes confidence calculators to assess the certainty of object distance values and positions, and a corrector to refine the distance image signal based on these confidences, enabling accurate determination of the depth range where objects exist.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If a large collation region is used to evaluate correlation, then more objects can be covered, but incorrect object distance calculations occur when multiple objects at different distances are included
Solution Approach 1:
The patent divides the collation region into multiple sub-regions and performs correlation evaluation separately in each sub-region. This segmentation allows the system to handle large areas while maintaining precision by evaluating each sub-region independently, thus avoiding the mixing of objects at different distances that occurs in a single large collation region.
Solution Approach 2:
The patent applies different processing strategies to different regions based on their characteristics. By evaluating correlation locally in each sub-region rather than uniformly across the entire collation region, the system adapts to local variations in object distance and density, improving overall measurement precision while covering large areas.
2Measurement precision
If correlation evaluation is performed with high precision requirements, then accurate object distance can be calculated, but incorrect evaluations occur when contrast changes little or noise is high
Solution Approach 1:
The patent introduces a confidence evaluation mechanism that assesses the reliability of correlation evaluation results. By calculating confidence values based on correlation coefficients and comparing them against thresholds, the system provides feedback to identify and correct incorrect evaluations, thereby improving both precision and reliability in challenging conditions such as low contrast or high noise.
Solution Approach 2:
The patent performs preliminary confidence evaluation before finalizing object distance calculations. By pre-assessing the quality of correlation evaluation and identifying unreliable results in advance, the system can apply corrective measures or alternative processing methods, ensuring that final distance calculations are based on reliable data even when contrast is low or noise is high.
3Measurement precision
If confidence evaluation is added to assess object distance certainty, then accuracy improves, but processing complexity increases
Solution Approach 1:
The patent implements a self-evaluating system where the correlation evaluation process automatically generates confidence values based on its own results. By using the correlation coefficient and its variation as intrinsic measures of confidence, the system performs self-diagnosis and quality assessment without requiring external complex verification mechanisms, thus improving accuracy while limiting complexity growth.
Data Source
AI summary
A distance information processing apparatus includes a memory and a processor that function as an acquirer configured to acquire a distance image signal constituted by a plurality of distances to an object in a depth direction in regions of an image signal. The memory and processor also function as a determiner configured to determine positional confidence of the distances for the regions of the image signal based on differences among a plurality of positions in an in-plane direction perpendicular to the depth direction, the plurality of positions respectively corresponding to the plurality of distance in the depth direction. The determiner is further configured to determine the positional confidence based on an image SN ratio of the image signal.