Image Sensor Pixel Change Detection for Time-Surface Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image sensors output redundant background information, leading to increased storage and computing demands in machine vision systems, and the asynchronous event streams from dynamic vision sensors are not compatible with mainstream algorithms, necessitating inefficient data conversion.
Innovation Solution
An image sensor with a pixel collection circuit that includes a time signal generation module and a peripheral reading unit to generate and output a time surface frame, reducing the need for data conversion by directly providing time-space information for machine vision algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If a conventional image sensor outputs entire images at fixed frame rate, then complete image information is provided, but redundant background information increases storage and computing burden
Solution Approach 1:
The patent extracts only the necessary dynamic information (change detection signals) from the complete image data, eliminating redundant background information while preserving essential object motion data. This extraction is achieved through per-pixel change detection circuits that only output signals when changes occur, rather than transmitting entire frames.
Solution Approach 2:
The system transitions from static fixed-frame-rate image output to dynamic asynchronous event streams that continuously update based on detected changes. This dynamic approach allows the system to adaptively transmit only necessary information at the moment of change, reducing overall data volume while maintaining complete information about object movements.
2Productivity
If a dynamic vision sensor outputs asynchronous event streams, then dynamic information is captured efficiently, but compatibility with mainstream machine vision algorithms is lost
Solution Approach 1:
The patent introduces a time signal generation module as an intermediary that converts asynchronous event streams into synchronous time surface frames. This intermediary layer bridges the gap between the efficient asynchronous detection capability and the compatibility requirements of mainstream algorithms, allowing both advantages to coexist.
Solution Approach 2:
The system changes the temporal parameter representation from asynchronous event-based timing to synchronous time surface frame timing. By generating time signals that synchronize with the readout clock, the system transforms the data format into something compatible with conventional algorithms while preserving the efficiency benefits of asynchronous detection.
3Adaptability or versatility
If asynchronous event streams are converted to synchronous frame data, then algorithm compatibility is achieved, but external storage space and computing capability requirements increase
Solution Approach 1:
The time signal generation module performs the conversion from asynchronous to synchronous format at the sensor level before data reaches external processing systems. This preliminary action eliminates the need for resource-intensive conversion in later stages, reducing overall storage and computing requirements while maintaining algorithm compatibility.
4Loss of information
If conventional image sensors are used, then complete image data is available, but dynamic information of interest is lost in static frame data
Solution Approach 1:
The patent segments the image data into individual pixel-level change detection events rather than processing entire frames. This segmentation allows the system to track dynamic information at the pixel level while discarding static background information, achieving both high dynamic information preservation and reduced data volume.
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
The image sensor reduces storage and computation overhead by directly outputting time surface frames, enhancing the efficiency of image processing systems in tasks like object identification and tracking.
Implementation Method 1
a photodetection unit operative to output a first electrical signal representing the intensity of the light signal that irradiates on the photodetection unit in real time
Data Source
Figure 1
Figure 2~3
Figure 4A~5
AI summary
The present disclosure provides a pixel collection circuit and an image sensor. The image collection circuit includes: a movement detection module configured to detect a change in a light intensity in a field of view, and generate a pixel triggering signal when the change in the light intensity exceeds a predetermined threshold so as to indicate the pixel collection circuit to be in a triggered state; and a time signal generation module coupled to the movement detection module and configured to generate a time signal upon the receipt of the pixel triggering signal, the time signal being used to represent time information about a time point at which the pixel collection circuit is triggered. The movement detection module and the time signal generation module are coupled to a reading unit, so that the reading unit scans the pixel collection circuit to output the time information about the pixel collection circuit.