Stereo Vision Height Map Object Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current object detection and tracking systems face challenges in robustness due to background clutter, lighting conditions, occlusion, and difficulty in distinguishing between objects of different heights, especially in monocular systems, and struggle to provide unobtrusive, low-maintenance, and precise traffic counting with the ability to differentiate between incoming and outgoing traffic.
Innovation Solution
A computer-implemented detection and tracking system using stereo cameras to generate height maps from image pairs, which allows for robust object detection and tracking by overcoming background clutter and lighting issues, and includes a counting mechanism to differentiate between entering and exiting objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If monocular video streams are used for object detection and tracking, then the system structure is simple, but the system suffers from background clutter, lighting condition sensitivity, and difficulty in distinguishing objects of different heights
Solution Approach 1:
The patent transitions from monocular 2D video streams to stereo 3D vision systems. By adding depth perception through stereoscopic cameras, the system generates height maps that provide three-dimensional information about objects in the scene. This dimensional enhancement allows the system to distinguish objects of different heights reliably and overcome background clutter that plagues monocular systems, while maintaining reasonable system complexity through integrated processing pipelines.
2Ease of manufacture
If background differencing is used in monocular systems, then the detection process is simple, but the presence of background clutters negatively affects robustness and reliability
Solution Approach 1:
The patent replaces 2D background differencing with 3D height map analysis. Instead of comparing intensity values in the image plane, the system compares height information from stereo vision. This dimensional transition eliminates the fundamental problem of background clutter affecting detection, as height information provides a new perspective that separates foreground objects from backgrounds regardless of their visual similarity in 2D space.
3Productivity
If adaptive template matching is used, then object detection can be performed, but detections tend to drift from true locations or get fixed to strong features in the background
Solution Approach 1:
The patent replaces 2D template matching with 3D height-based object identification. By using height maps derived from stereo vision, the system matches objects based on their three-dimensional shape and elevation rather than two-dimensional appearance. This eliminates the drift problem and false locking to background features, as height information provides unique geometric signatures that are independent of lighting conditions and background textures.
4Reliability
If stereo systems are used to generate height maps, then robustness against lighting conditions is improved, but the system complexity increases
Solution Approach 1:
The patent combines multiple stereo camera systems into an integrated setup with unified processing. By merging the calibration, height map generation, and object detection processes into a cohesive system architecture, the patent achieves robust lighting-independent operation while managing system complexity through standardized interfaces and coordinated processing pipelines that treat the stereo systems as a unified whole rather than separate components.
5Productivity
If range background differencing is used in stereo systems, then object detection can be performed, but difficulty exists in differentiating between multiple closely positioned objects
Solution Approach 1:
The patent enhances 3D height map analysis to resolve closely positioned objects. By utilizing the full three-dimensional information from stereo vision, including depth, height, and spatial relationships, the system can distinguish between objects that are close together in 2D space but separated in 3D space. The height dimension provides an additional separation criterion that enables precise differentiation of closely positioned objects that would be indistinguishable in monocular systems.
Data Source
AI summary
A method of identifying, tracking, and counting human objects of interest based upon at least one pair of stereo image frames taken by at least one image capturing device, comprising the steps of: obtaining said stereo image frames and converting each said stereo image frame to a rectified image frame using calibration data obtained for said at least one image capturing device; generating a disparity map based upon a pair of said rectified image frames; generating a depth map based upon said disparity map and said calibration data; identifying the presence or absence of said objects of interest from said depth map and comparing each of said objects of interest to existing tracks comprising previously identified objects of interest; for each said presence of an object of interest, adding said object of interest to one of said existing tracks if said object of interest matches said one existing track, or creating a new track comprising said object of interest if said object of interest does not match any of said existing tracks; updating each said existing track; and maintaining a count of said objects of interest in a given time period based upon said existing tracks created or modified during said given time period.