Selective Pixel Extraction for Real-Time Object Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D reconstruction techniques face challenges in real-time processing on embedded chips with low throughput, leading to precision issues such as non-detection of objects or features, especially near edges, due to the high computational demands of structure from motion algorithms.
Innovation Solution
An information processing device that includes a pixel extraction unit, self-location estimation unit, and three-dimensional location estimation unit, which selects specific pixel groups for image analysis, reduces processor load by minimizing pixel usage, and employs image correction to enhance accuracy, allowing real-time high-precision 3D reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If structure from motion algorithms are applied to perform high-precision 3D reconstruction, then measurement precision is improved, but productivity deteriorates due to high computational demands
Solution Approach 1:
The patent segments the image processing task by extracting only specific pixel groups (e.g., edges, corners, high-gradient regions) rather than processing all pixels. This segmentation allows the system to maintain high 3D reconstruction precision for critical features while significantly reducing the overall computational load and improving processing throughput on embedded devices
Solution Approach 2:
The patent applies local quality by treating different regions of the image differently - high-precision reconstruction is applied locally to regions with edges and high gradients where features are critical, while lower precision or no processing is applied to uniform regions. This selective approach maintains measurement precision where needed while improving overall productivity
2Productivity
If the number of points utilized in reconstruction processing is reduced, then productivity is improved, but measurement precision deteriorates due to non-detection of objects or features
Solution Approach 1:
The patent extracts only the most informative pixel groups (edges, corners, high-gradient points) from the entire image. By taking out only these critical points rather than processing all pixels or using a random sample, the system achieves both high productivity and maintains measurement precision, as the extracted points are specifically those most likely to contain detectable features
Solution Approach 2:
The patent performs preliminary image processing to identify and extract pixel groups with edges and high gradients before the main reconstruction processing. This preliminary action filters and selects only the most promising points for 3D reconstruction, ensuring that the reduced set of points still contains sufficient information for accurate object and feature detection
Data Source
AI summary
According to one embodiment, an information processing device includes a pixel extraction unit, a self-location estimation unit, and a three-dimensional location estimation unit. The pixel extraction unit acquires information from a plurality of pixels in an image, the plurality of pixels having a shift of a first number of pixels in a first direction, and having a shift of a second number of pixels in a second direction crossing the first direction along with a shift of at least one pixel in the first direction. The self-location estimation unit estimates a location and orientation of the sensor providing the image. The three-dimensional location estimation unit reconstructs three-dimensional information based on the information from the plurality of pixels extracted by the pixel extraction unit and the estimated location and orientation of the sensor.