Event-Based Stereo Traffic Monitoring for Low-Data Night Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing traffic monitoring devices, particularly those using stereo cameras with conventional image sensors, face challenges such as high data volume generation, dependence on lighting conditions, and inefficiency in capturing dynamic scenes, making reliable traffic monitoring difficult, especially at night, and are limited in their ability to detect multiple vehicles and traffic violations beyond speed.
Innovation Solution
A traffic monitoring device utilizing event-based image sensors with pixel matrices that independently detect relative changes in light intensity, allowing for efficient detection of dynamic elements while reducing data volume and energy consumption, and enabling reliable monitoring regardless of lighting conditions, with the capability to detect multiple vehicles and violations like speed, distance, and vehicle class.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If conventional image sensors (CMOS or CCD) are used in stereo cameras for traffic monitoring, then complete scene coverage is achieved, but very large amounts of data are generated and energy consumption increases
Solution Approach 1:
The patent extracts only the dynamically changing elements from the complete scene by using event-based image sensors that detect relative changes in light intensity. Instead of capturing the entire static scene, the system extracts and transmits only the events (changes) occurring in the monitored area, thereby reducing data volume while preserving all relevant traffic information.
Solution Approach 2:
The patent transitions from static frame-based capture to dynamic event-based capture. The event-based image sensors continuously monitor for changes and generate data only when dynamic events occur, adapting the data generation rate to the actual traffic activity level rather than operating at a fixed high rate regardless of scene content.
2Loss of information
If conventional image sensors are used in stereo cameras for traffic monitoring, then complete scene coverage is achieved, but energy consumption increases
Solution Approach 1:
The patent extracts only the dynamically changing elements from the complete scene by using event-based image sensors that detect relative changes in light intensity. Instead of capturing the entire static scene, the system extracts and transmits only the events (changes) occurring in the monitored area, thereby reducing data volume while preserving all relevant traffic information.
Solution Approach 2:
The patent transitions from static frame-based capture to dynamic event-based capture. The event-based image sensors continuously monitor for changes and generate data only when dynamic events occur, adapting the data generation rate to the actual traffic activity level rather than operating at a fixed high rate regardless of scene content.
3Productivity
If conventional image sensors are used in stereo cameras for traffic monitoring, then frame-based capture is achieved, but reliability at night is compromised due to dependence on lighting conditions
Solution Approach 1:
The patent changes the fundamental operating parameter of the image sensors from fixed exposure time capture to variable threshold-based event detection. The event-based sensors detect relative changes in light intensity regardless of absolute lighting levels, allowing consistent performance across varying lighting conditions including nighttime without requiring exposure time adjustment.
4Measurement precision
If stereo cameras with conventional image sensors are used for traffic monitoring, then speed detection is achieved, but detection of multiple traffic violations is limited
Solution Approach 1:
The patent makes the traffic monitoring system universal by enabling detection of multiple types of violations beyond speed. The event-based stereo camera system can detect speed violations, distance violations (tailgating), red light running, and other traffic infractions by analyzing the spatio-temporal patterns of detected events, allowing a single system to perform multiple monitoring functions.
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 device achieves efficient and reliable traffic monitoring by capturing only dynamic elements, reducing data volume and energy consumption, and providing high dynamic range, while enabling simultaneous detection of multiple vehicles and compliance with various traffic regulations, including speed limits and minimum distances, independent of lighting conditions.
Implementation Method 1
the pixels are each designed to independently and asynchronously detect relative changes in light intensity as events
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
The invention relates to a traffic monitoring device (10a; 10b) for detecting vehicles (12a, 14a, 16a, 18a, 20a; 12b, 14b) on a roadway (22a; 22b), with a detection unit (24a; 24b) comprising at least one stereo camera (26a; 26b) with two image sensors (32a, 34a; 32b, 34b) arranged along a baseline (28a; 28b) at a predefined distance (30a; 30b) from each other, which are directed at at least partially overlapping sections of the roadway (22a; 22b), and with an evaluation unit (36a) for evaluating data acquired by the image sensors (32a, 34a; 32b, 34b).To increase efficiency, it is proposed that the image sensors (32a, 34a; 32b, 34b) of the stereo camera (26a; 26b) are configured as event-based image sensors (32a, 34a; 32b, 34b) and each comprise a pixel matrix (38a) with a plurality of pixels (40a, 42a), wherein the pixels (40a, 42a) are each configured to detect relative changes in light intensity independently and asynchronously as events (44a, 46a), wherein the detection unit (24a; 24b) is configured to detect events (44a, 46a, 44a', 46a') occurring simultaneously at identical object points in an overlap area (48a; 48b) of the subsections using both event-based image sensors (32a, 34a; 32b, 34b) to capture the stereo camera (26a; 26b) and to provide it as corresponding events (44a, 46a, 44a', 46a') of an event data set (50a, 52a).