Coordinate Transformation Matrix Selection for Video Mapping

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional techniques fail to accurately map video coordinates to real-world coordinates due to distortions like lens distortion, uneven scaling, and shearing, and are inadequate for detecting overlapping views and blind spots between cameras.

Innovation Solution

A processor-implemented method and system that obtains images from capture devices, identifies objects, determines image-frame coordinates, selects appropriate coordinate transformation matrices, and calculates real-world coordinates using these matrices to track object trajectories across multiple camera views.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional coordinate transformation techniques are used, then the mapping process is simple, but the mapping accuracy decreases due to lens distortion, uneven scaling, and shearing

Engineering Contradiction:
Improvemapping accuracyVSAvoidcoordinate transformation complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the video frame into multiple regions, each with its own coordinate transformation matrix. This segmentation allows each region to be mapped accurately to the floor plan without being affected by distortions in other regions, thereby improving overall mapping accuracy while managing complexity through localized transformations.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies different coordinate transformation matrices to different regions of the video frame based on their specific distortion characteristics. Each region is transformed using a matrix optimized for its local properties, ensuring high accuracy in each area while accommodating variations in lens distortion, scaling, and shearing across the frame.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If multiple coordinate transformation matrices are used for different regions, then mapping accuracy improves, but the complexity of selecting the appropriate matrix increases

Engineering Contradiction:
Improvecoordinate transformation accuracyVSAvoidmatrix selection difficulty
Core Design Contradiction:
Measurement precisionVSDifficulty of detecting and measuring

Solution Approach 1:

The patent pre-calculates and stores multiple coordinate transformation matrices for different regions of the video frame before actual mapping occurs. This preliminary preparation allows the system to quickly retrieve and apply the appropriate matrix during real-time operation without performing complex calculations at that moment, thereby reducing the difficulty of matrix selection during execution.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary mechanism that automatically selects the appropriate coordinate transformation matrix based on the object's position in the video frame. This intermediary layer simplifies the selection process by using the object's coordinates to determine which pre-calculated matrix to apply, eliminating the need for manual or complex automated matrix selection.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Measurement precision

If conventional techniques are used to detect overlapping views and blind spots, then the detection process is straightforward, but the detection accuracy is insufficient

Engineering Contradiction:
Improveoverlapping view detection accuracyVSAvoiddetection system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges the coordinate transformation process with the overlapping view and blind spot detection process. By transforming video coordinates to floor plan coordinates using multiple region-specific matrices, the system can accurately determine which areas are covered by which cameras and identify overlaps and blind spots through the transformed coordinate mappings, improving detection accuracy while integrating functions rather than adding separate complex systems.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP3051495B1Systems and methods for mapping object coordinates from a video frame view to real world coordinates
Publication Date: 2018.09.19 WIPRO LTD
  • EP3051495B1 patent drawingFigure 1
  • EP3051495B1 patent drawingFigure 2
  • EP3051495B1 patent drawingFigure 3

AI summary

A method for mapping object coordinates from a video frame view to real world coordinates includes obtaining an image from an image capture device and identifying, via one or more hardware processors, an object depicted in the image. Image-frame object coordinates for the object are determined, and one of a plurality of coordinate transformation matrices associated with the image capture device is selected, based on the image-frame object coordinates for the object. Real-world object coordinates for the object are calculated using the image-frame coordinates and the selected coordinate transformation matrix, and a trajectory of the object is determined using the calculated real-world object coordinates.