Event-Driven Photoarray Reduces Data Volume
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Real-time artificial vision using photoarrays is limited by frame rate and generates excessive redundant data, requiring costly post-processing.
Innovation Solution
A photoarray with a topologically one- or two-dimensional array of cells, each equipped with a photosensor, a capacitor charged by the time derivative of the sensor signal, threshold detectors, and a discharge device to reduce data by only transmitting changes in light intensity, along with a signal collector to manage output signals and reset capacitors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional photoarrays sample at a fixed frame rate, then real-time vision is achieved, but a huge amount of redundant data is generated requiring costly post processing
Solution Approach 1:
The patent extracts only the essential information (changes in light intensity) from the full frame data by using event-driven sampling. Each pixel outputs a signal only when the light intensity changes, effectively taking out the relevant data while discarding the redundant information that would otherwise require expensive post-processing to filter.
Solution Approach 2:
The patent implements dynamic sampling where the sampling rate of each pixel is automatically adjusted based on the activity level in that region. Active regions with changing light intensity are sampled at high rates, while static regions are sampled minimally or not at all, creating a dynamic data generation process that adapts to the scene content.
2Speed
If all pixels are sampled continuously at high frame rates, then real-time detection is achieved, but the data processing burden increases significantly
Solution Approach 1:
The system dynamically adjusts the sampling frequency of each pixel based on its activity level. When a pixel detects a change in light intensity, it generates an event signal immediately, providing fast response. When the scene is static, pixels remain dormant, dramatically reducing the overall data processing burden while maintaining real-time responsiveness where needed.
Solution Approach 2:
Different regions of the image sensor have different sampling characteristics based on local activity. Regions with motion or changing illumination generate events and are processed in real-time, while static regions contribute minimal data. This local differentiation allows the system to achieve real-time performance for important events without the computational burden of processing all pixels at full frame rate.
3Productivity
If the photoarray transmits only changed data, then data transfer efficiency is improved, but the system must detect and encode changes accurately
Solution Approach 1:
Each pixel incorporates a feedback mechanism that compares the current light intensity with the previous state. When a change is detected, the pixel generates an event signal that includes information about the nature of the change. This feedback-based change detection ensures accurate measurement of light intensity variations while enabling efficient event-driven data transmission.
Solution Approach 2:
The patent replaces traditional continuous analog-to-digital conversion with an event-driven digital output mechanism. Instead of continuously converting and transmitting analog signals, the system uses threshold-based detection that triggers digital event signals only when changes occur, substituting a simpler detection mechanism for the more complex continuous conversion process while maintaining measurement accuracy for changes.
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 approach significantly reduces data processing needs and allows for higher data transfer rates by only communicating changes in light intensity, enabling more efficient real-time artificial vision.
Implementation Method 1
Each cell has a photodiode D generating a current signal dependent on the light intensity at the cell
Data Source
Figure 1~2
Figure 3
AI summary
The photoarray comprises a one- or two- dimensional array of cells (10), where each cell (10) has a photosensor (D, Tl - T4) generating a sensor signal dependent on a light intensity at the cell (10), a first capacitor (Cl) being charged by the time-derivative of the current, at least one threshold detector (T9 - TIl; T12 - T14) detecting if the voltage over the first capacitor (Cl) exceeds a threshold value and generating an output signal if yes, and a discharge device (T7) for discharging the first capacitor after occurrence of said output signal. Such a cell is generates an event only when the incoming light intensity changes, which reduces the amount of data to be processed from the photoarray.