Image Recognition Device Using Multi-Sensitivity Pixel Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image recognition technologies face accuracy deterioration due to artifacts in high dynamic range (HDR) image combinations, which affect subject recognition.
Innovation Solution
An image recognition device captures multiple images at the same exposure start timing using imaging pixels with different sensitivities to generate image data, allowing for subject recognition from high-sensitivity and low-sensitivity images before HDR combination, thereby eliminating artifact influence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If HDR image combination is used to capture images with different sensitivities, then dynamic range is improved, but subject recognition accuracy deteriorates due to artifacts
Solution Approach 1:
The imaging pixels are segmented into multiple types with different sensitivities (e.g., first imaging pixels with first sensitivity, second imaging pixels with second sensitivity). Each pixel type independently captures images, allowing separate processing that avoids HDR combination artifacts while maintaining high dynamic range capability through selective use of different pixel types based on lighting conditions.
Solution Approach 2:
Instead of combining images from different sensitivities to achieve HDR (which causes artifacts), the invention inverts the approach by using images from different sensitivity pixels separately for recognition. The system selects which pixel type's images to use based on the subject's brightness characteristics, eliminating the need for problematic HDR combination while preserving dynamic range benefits.
2Measurement precision
If multiple images with different sensitivities are captured and processed separately, then subject recognition accuracy is improved, but device complexity increases
Solution Approach 1:
Different pixel types are assigned to different local regions or have different sensitivity characteristics optimized for specific conditions. The control unit selectively activates appropriate pixel types based on local image characteristics (e.g., using first imaging pixels for bright subjects, second imaging pixels for dark subjects), simplifying processing by avoiding unnecessary computations while maintaining high recognition accuracy.
Solution Approach 2:
The system changes operational parameters (which pixel type is active) based on detected conditions. By adjusting the sensitivity parameter of the active imaging pixels according to subject brightness, the system achieves accurate recognition across varying conditions without requiring complex processing of all pixel types simultaneously, thus reducing overall device complexity.
Data Source
AI summary
An image recognition device (image recognition system 100) according to the present disclosure includes an imaging unit (10) and a recognition unit (14). The imaging unit (10) captures a plurality of images at the same exposure start timing in one frame period by using imaging pixels having different sensitivities to generate image data. The recognition unit (14) recognizes a subject from each of the image data. The imaging unit (10) includes a pixel array in which a plurality of imaging pixels having different exposure times, different light transmittances of color filters, or different light receiving areas are two-dimensionally arranged.