Video Surveillance Coordinate Transformation Using Camera Orientation Parameters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video surveillance systems face inefficiencies and inaccuracies in transforming image coordinates to map coordinates due to computationally difficult camera calibration procedures and inaccuracies in linear interpolation techniques.
Innovation Solution
A method and system that select a reference point with known image and map coordinates, compute transformation parameters including rotation and tilt angles of the camera, and use these parameters to accurately determine map coordinates of a target location within an image, reducing the need for complex camera calibration and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If 4-point or 9-point linear interpolation is used to derive map coordinates from image coordinates, then the transformation process is simple, but the map coordinates become inaccurate
Solution Approach 1:
The patent changes the transformation parameters from simple linear interpolation coefficients to comprehensive parameters including rotation angles, tilt angles, and focal length. This allows the system to account for camera orientation and optical characteristics, significantly improving map coordinate accuracy while maintaining computational feasibility through a standardized transformation model.
2Productivity
If conventional linear interpolation techniques are used, then the computation is simpler, but camera calibration procedures become computationally difficult
Solution Approach 1:
The patent performs preliminary camera calibration to determine intrinsic parameters (focal length, principal point) and extrinsic parameters (rotation angles, tilt angles) before the actual coordinate transformation. This preliminary action separates the complex calibration process from the real-time transformation, making the calibration done once beforehand and the transformation computationally efficient during operation.
Solution Approach 2:
The patent transforms the calibration problem into determining a set of physical parameters (rotation angles, tilt angles, focal length) that describe the camera's orientation and optical properties. These parameters are then used in a standardized transformation model, converting a computationally difficult calibration procedure into a more manageable parameter estimation problem.
3Device complexity
If a single reference point is used for transformation, then the calibration process is simplified, but the system needs accurate camera parameters
Solution Approach 1:
The patent uses a single reference point with known map coordinates to solve for multiple camera parameters simultaneously (rotation angles, tilt angles, and focal length). By changing from using multiple reference points to using one reference point with a comprehensive parameter model, the system simplifies the calibration process while maintaining accuracy through the physically-based transformation model that accounts for camera orientation and optics.
Data Source
AI summary
Systems and methods for transformations between image and map coordinates, such as those associated with a video surveillance system, are described herein. An example of a method described herein includes selecting a reference point within the image with known image coordinates and map coordinates, computing at least one transformation parameter with respect to a location and a height of the camera and the reference point, detecting a target location to be tracked within the image, determining image coordinates of the target location, and computing map coordinates of the target location based on the image coordinates of the target location and the at least one transformation parameter.