Event Token Camera Sensor for High Dynamic Range Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cameras face challenges in achieving high dynamic range, frameless image capture, and efficient in-pixel processing due to limitations in integrator capacity, analog to digital conversion, and noise susceptibility, which results in compromised image quality and saturation issues.
Innovation Solution
A camera sensor with event token based image capture and reconstruction, featuring a photodetector, integrator, in-pixel processor, and communication pipeline that generates and transmits event tokens to post-processing circuitry, allowing for asynchronous processing and accurate image intensity calculation independent of integrator capacity and ADC sensitivity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the integration rate is increased to improve quality in darker images, then measurement precision is improved, but the integrator saturates quicker in lighter images causing loss of information
Solution Approach 1:
The integrator capacity is made dynamically adjustable through multiple capacitive elements (first, second, and third integrators with different capacities) that can be selectively activated. This allows the system to adapt integrator size to scene brightness conditions, using smaller integrators for bright scenes to prevent saturation and larger integrators for dark scenes to improve signal-to-noise ratio.
Solution Approach 2:
The system changes the integration capacity parameter based on scene conditions and pixel type. Different pixel types (first, second, and third photodetector types) are assigned different integrator capacities, allowing the integration parameter to be optimized for specific imaging scenarios without compromising overall system performance.
2Loss of information
If a larger integrator is used to prevent saturation in bright images, then loss of information is reduced, but measurement precision deteriorates in dark regions
Solution Approach 1:
The integrator capacity is made dynamically adjustable through multiple capacitive elements (first, second, and third integrators with different capacities) that can be selectively activated. This allows the system to adapt integrator size to scene brightness conditions, using smaller integrators for bright scenes to prevent saturation and larger integrators for dark scenes to improve signal-to-noise ratio.
Solution Approach 2:
The system changes the integration capacity parameter based on scene conditions and pixel type. Different pixel types (first, second, and third photodetector types) are assigned different integrator capacities, allowing the integration parameter to be optimized for specific imaging scenarios without compromising overall system performance.
3Productivity
If ADC is placed adjacent to each pixel for in-pixel processing, then processing speed is improved, but chip area for photodetectors is reduced
Solution Approach 1:
The ADC function is extracted from the pixel array and implemented as shared column-parallel readout circuits. Instead of having ADCs adjacent to each pixel, the patent uses a time-division multiplexed approach where a single ADC in each column services multiple pixels sequentially, eliminating the need for per-pixel ADCs and freeing up chip area for larger photodetectors.
Solution Approach 2:
The readout circuitry is designed to be universal and multi-functional, serving multiple pixels through time-division multiplexing. The same ADC and readout circuitry process signals from different pixels at different times, providing in-pixel processing capabilities without requiring dedicated hardware for each pixel.
4Device complexity
If analog signals are sent to post-processing circuitry to reduce chip area, then device complexity is reduced, but signal quality deteriorates due to noise and crosstalk
Solution Approach 1:
The patent replaces the analog signal transmission system with a digital system. Event tokens are generated digitally based on integrator overflow conditions and transmitted through digital communication pipelines to post-processing circuitry. This substitution of digital for analog transmission eliminates susceptibility to noise and crosstalk while maintaining area efficiency.
Solution Approach 2:
Event tokens serve as intermediaries between the photodetector array and post-processing circuitry. Instead of directly transmitting analog or digital pixel values, the system uses event tokens that encode timing and intensity information, providing robust noise-immune communication while reducing the bandwidth requirements of the transmission pipeline.
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
This solution enhances image accuracy and dynamic range, reduces noise, and allows for frameless capture and selective region enhancement, achieving finer gradations in image reproduction and lower area costs at the pixel sensor.
Implementation Method 1
a photodetector for capturing light from a portion of a scene and for producing a signal indicative of the light
Data Source
AI summary
The subject matter described herein includes a camera sensor with event token based image capture and reconstruction. The sensor includes a photodetector for capturing light from a portion of a scene and for producing a signal indicative of the light. An integrator is coupled to the photodetector for accumulating charge resulting from the signal output by the photodetector and can be reset each time the charge reaches a predetermined level. An in-pixel processor is coupled to the integrator for resetting the integrator and generating an event token each time the predetermined level of charge is accumulated. A communication pipeline communicates the event tokens for downstream processing. A postprocessor is coupled to the pipeline for receiving the event tokens and for determining output intensity for the portion of the scene being reconstructed based on a number of reset events and a time between at least two of the event tokens.