Query-Driven Image Sensing for Low Power IoT
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional CMOS imagers have high pixel density but high power consumption due to frame scanning, while silicon retina chips with asynchronous event-driven methods have low pixel density and high power usage, making them inefficient for energy and data rate in applications like IoT and micro sensors.
Innovation Solution
A query-driven imaging approach using clocked time-division multiplexing to scan pixels for threshold changes, eliminating static idle power and event-handling overhead, resulting in reduced data rate and power consumption by only transmitting pixels with intensity changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of moving object
If frame scanning is used to synchronize and stream pixel intensity data, then pixel density can be high, but data rate and power consumption become excessively high
Solution Approach 1:
The patent extracts only the essential information (pixels with intensity changes exceeding threshold) from the complete frame data, eliminating redundant transmission of static background pixels. This selective extraction reduces data rate and power consumption while maintaining high pixel density in the sensor array.
Solution Approach 2:
The patent changes the parameter being transmitted from absolute pixel intensity values to relative temporal contrast signals (event data indicating intensity changes). This parameter transformation reduces the amount of data that needs to be transmitted and processed, thereby reducing power consumption while preserving high spatial resolution.
2Use of energy by moving object
If asynchronous event-driven methodology is used to stream events only when they occur, then data rate is reduced, but pixel density decreases and static power overhead increases
Solution Approach 1:
The patent merges the advantages of frame-based scanning (high pixel density, simple architecture) with event-driven methods (reduced data rate). By using a scan-driven approach that queries pixels for events and transmits only changed pixels with temporal contrast information, it achieves low data rates without requiring complex per-pixel event handling circuitry, thus maintaining high pixel density.
3Speed
If pixels continuously monitor and rapidly route events through asynchronous protocol, then event detection is fast, but power consumption and area increase due to static current and overhead circuitry
Solution Approach 1:
Instead of continuous monitoring, the patent uses periodic scanning to query pixels for events. The scan-driven circuitry periodically checks pixel states and transmits data only when changes are detected. This periodic action reduces power consumption compared to continuous monitoring while maintaining event detection capability through the temporal contrast calculation.
4Loss of information
If all pixel data is transmitted for every frame, then complete image information is available, but data rate becomes excessively high and most data is redundant
Solution Approach 1:
The patent extracts only the meaningful changes in the image data by comparing current pixel intensity with previous frame values and transmitting only those pixels where the intensity change exceeds a threshold. This extraction of essential information maintains image completeness for dynamic elements while eliminating redundant static background data, reducing data rate significantly.
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 power consumption by 10× compared to state-of-the-art Dynamic Vision Sensors, achieving efficient energy use and data streaming suitable for applications like security surveillance and drone navigation.
Implementation Method 1
each pixel configured for generating a pixel signal corresponding to an intensity of the detected light impinging thereon
Data Source
AI summary
The multimodal query-driven imager provides efficient coding and streaming of visual information acquired directly on the focal plane. The query-driven approach to visual event coding uses clocked time-division multiplexing to continuously scan the array, querying each pixel for threshold change events in pixel intensity.