Stereo Height-Map Traffic Sensing for Building Energy Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current object detection and tracking systems for human objects in facilities face challenges such as background clutter, occlusion, lighting conditions, and the inability to distinguish between incoming and outgoing traffic, as well as between different heights, leading to unreliable and costly maintenance requirements.
Innovation Solution
A computer-implemented system using stereo cameras to capture images, convert them into height maps, and track human objects by detecting local maxima in the height maps, allowing for precise counting and differentiation between moving and static objects, while also integrating RFID technology to separately track employees and customers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If monocular video streams are used for object detection and tracking, then the system is simpler and cheaper, but the system becomes unreliable under adverse lighting conditions and background clutter
Solution Approach 1:
The patent transitions from monocular 2D intensity images to stereo 3D range/height maps, adding a depth dimension. This dimensional change allows the system to detect objects based on their spatial position and height rather than intensity, making it immune to lighting conditions, shadows, and background clutter that plague monocular systems.
2Reliability
If stereo systems are used for robust detection, then reliability improves under adverse conditions, but the system complexity and cost increase
Solution Approach 1:
The patent extracts and utilizes only the height map information from stereo images, discarding the intensity information that causes problems with lighting and background clutter. By focusing on the essential height dimension and applying background differencing only to height maps, the system achieves robustness without requiring complex processing of all image data.
Solution Approach 2:
The patent applies different processing strategies to different types of information: intensity images are used for initial object detection, while height maps are used for robust tracking and background differencing. Each data type is processed according to its strengths, with height maps handling the demanding task of reliable detection under varying lighting conditions.
3Ease of operation
If background differencing is used in monocular systems, then object detection is simplified, but the system becomes susceptible to background clutter and lighting changes
Solution Approach 1:
The patent applies background differencing not to intensity images but to height maps derived from stereo vision. Since height information is unaffected by lighting conditions and background clutter, the background differencing operation becomes robust while maintaining its simplicity. Objects are detected by finding regions where the height map differs from the background height map.
4Adaptability or versatility
If adaptive template matching is used for object detection, then detection can adapt to changing conditions, but detections drift from true locations and are prone to occlusion
Solution Approach 1:
The patent performs template matching and object recognition in the height map domain rather than intensity domain. Height maps provide accurate spatial information that is not affected by lighting variations, allowing templates to be matched precisely to object locations even when occluded or when lighting conditions change. The height dimension provides a stable reference for accurate localization.
Data Source
AI summary
A method of managing energy consumption in an enclosed spaced includes the step of providing temperature and foot traffic data to a plan generator. The method further includes the steps of generating an energy plan based on the temperature and foot traffic data and controlling one or more energy consuming devices based on the energy plan.