Event Sensor Pixel-Group Readout for High Time Resolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing high-speed image sensors with low power consumption, such as event-based sensors, suffer from signal readout delays and time resolution artifacts due to row-based signal processing, leading to distorted object perception and potential loss of preceding signals if changes occur before readout is complete.
Innovation Solution
A photoelectric conversion device with a two-dimensional array of pixels and calculators, where each pixel group outputs spiking signals directly to corresponding calculators for immediate calculation, eliminating the need for row-based readout mechanisms and enabling asynchronous processing of event signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by stationary object
If row-based signal readout is used, then power consumption is reduced, but signal processing delay increases and time resolution deteriorates
Solution Approach 1:
The sensor is divided into multiple independent readout channels, each capable of processing signals from a specific region or set of pixels autonomously. This segmentation allows parallel signal processing across multiple channels, eliminating the sequential row-based readout bottleneck while maintaining low power consumption through selective activation of only necessary channels.
Solution Approach 2:
The patent transitions from traditional two-dimensional pixel arrays to a three-dimensional stacked architecture where photodetectors, readout circuits, and processing units are arranged in multiple layers. This vertical stacking enables simultaneous signal processing in the third dimension, allowing multiple pixels to be read out in parallel without increasing planar complexity, thus reducing processing delay while maintaining power efficiency.
2Device complexity
If row-based signal readout is used, then device complexity is reduced, but measurement precision deteriorates due to time resolution loss
Solution Approach 1:
The readout system is segmented into multiple independent channels, each with its own timing and processing resources. This allows simultaneous high-precision time stamping of events from different spatial regions without the sequential processing delays inherent in row-based readout, thereby maintaining measurement precision while managing device complexity through modular channel design.
Solution Approach 2:
Each pixel or pixel group is equipped with local event detection and time-stamping circuitry that operates autonomously without requiring centralized readout control. This self-service capability allows immediate local processing of events with high time resolution, eliminating the need for complex centralized arbitration and row-by-row scanning mechanisms.
3Productivity
If high-speed readout is implemented, then productivity is improved, but power consumption increases
Solution Approach 1:
The readout system operates in a periodic event-driven manner, activating high-speed readout channels only when events occur in specific regions, rather than continuously scanning all rows. This periodic activation maintains high productivity when needed while significantly reducing average power consumption compared to continuous high-speed readout of the entire sensor array.
Solution Approach 2:
The readout architecture dynamically allocates processing resources based on event distribution and priority. When events are concentrated in specific regions, only those regions' readout channels are activated at high speed, while other channels remain in low-power standby mode. This dynamic resource allocation achieves high productivity for event-rich regions without the prohibitive power cost of activating the entire sensor array.
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 configuration reduces signal processing delays, maintains high time resolution, and lowers power consumption by processing signals without bandwidth limitations, allowing accurate detection of moving objects even at high speeds.
Implementation Method 1
a photodiode that performs photoelectric conversion on incident light
Data Source
AI summary
A photoelectric conversion device is provided. The device includes a plurality of pixels each including a photoelectric conversion element, a plurality of calculators, and a processor. The plurality of pixels and the plurality of calculators are respectively arranged in a two-dimensional array. For the plurality of pixels, each pixel group of pixel groups composed of not less than two pixels of the plurality of pixels is connected to a corresponding calculator of the plurality of calculators, and each pixel group outputs a spiking signal generated by a pixel in a pixel group of the pixel groups to the corresponding calculator of the plurality of calculators. Each of the plurality of calculators executes calculation for the spiking signal, and the processor processes a calculation result input from each of the plurality of calculators.