Photoelectric Conversion Apparatus for Rapid In-Vehicle Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In-vehicle cameras face challenges with recognition processing delays and subject blur due to long accumulation periods, especially under low light conditions and high-speed object movement, which hinder prompt recognition and increase flicker issues.
Innovation Solution
A photoelectric conversion apparatus with a stacked structure comprising a sensor unit, counter, memory, and recognition unit that performs image signal processing and recognition based on the difference in count values at the start and end of accumulation periods, allowing for shorter accumulation periods and rapid recognition processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If the accumulation period is lengthened to capture images brightly under low light conditions, then image brightness is improved, but subject blur is generated in objects that move at high speed and recognition rate decreases
Solution Approach 1:
The patent divides the full frame period into multiple accumulation periods (e.g., 4 sub-frames), allowing recognition processing to be performed at multiple time points rather than waiting for a complete long accumulation period. This segmentation enables earlier recognition while using shorter accumulation periods, reducing subject blur while maintaining image brightness.
Solution Approach 2:
The patent performs recognition processing on intermediate accumulation results during the full frame period, rather than waiting for the complete accumulation period to end. This preliminary action allows recognition to occur at earlier time points (e.g., at 1/4, 1/2, or 3/4 of the full frame period), reducing latency and subject blur while the sensor continues accumulating photons for the remaining period.
2Reliability
If the accumulation period is set to at least 11 ms to suppress flicker caused by traffic signals, then flicker suppression is improved, but subject blur is generated and recognition processing cannot be performed promptly
Solution Approach 1:
The patent segments the accumulation period into multiple intervals, performing recognition processing at each segment boundary. This allows recognition to occur multiple times during what would traditionally be a single long accumulation period, reducing the effective recognition delay while maintaining the total accumulation duration needed for flicker suppression.
Solution Approach 2:
The patent maintains continuous photon accumulation throughout the full frame period while performing intermediate recognition processing. The accumulation continues uninterrupted even as recognition is performed on partial results, ensuring that the full accumulation period contributes to image brightness while enabling earlier recognition outputs.
3Ease of manufacture
If recognition processing is applied frame by frame under normal sensor driving at 30 fps, then processing simplicity is maintained, but recognition processing can only be executed every 33.3 ms causing delays when objects enter frames immediately after frame change
Solution Approach 1:
The patent divides each full frame period into multiple sub-frames with intermediate recognition processing points. This creates multiple recognition opportunities within what would traditionally be a single frame cycle, reducing the effective recognition delay from 33.3 ms to potentially 8.3 ms or less per sub-frame, while maintaining systematic processing.
Solution Approach 2:
The patent performs recognition processing on partial accumulation results before the full frame period completes. This preliminary recognition action provides earlier detection of objects that enter frames between traditional frame boundaries, reducing latency while the sensor continues accumulating photons for the remaining accumulation period.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables rapid recognition of obstacles and improved image quality by performing recognition every ¼ full frame period, reducing subject blur and latency in recognition results, facilitating timely vehicle control and display updates.
Implementation Method 1
a sensor unit configured to generate pulses corresponding to photons
Data Source
AI summary
A photoelectric conversion apparatus comprises a plurality of pixels, each provided with a sensor unit configured to generate pulses corresponding to photons, a counter that counts the number of pulses, and a memory that stores a count value of the counter, a recognition unit configured to perform recognition processing on an image signal generated based on the difference in the count values of the counter at a start time and an end time of an accumulation period, and a control unit configured to output the results of the recognition processing.