Image Perspective Conversion via Polygonal Sub-Region Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional image monitoring devices require extensive processing time and large memory capacity for pixel-by-pixel coordinate calibration of high-definition images, especially when the camera position, height, and view angle result in tilted images, leading to inefficient image calibration and storage.

Innovation Solution

An image perspective conversion method that partitions images into polygonal sub-regions, converts their coordinates into a world coordinate system, and interpolates pixels to generate a new image with a different angle of view, utilizing a processor and memory to manage the process efficiently.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If pixel-by-pixel coordinate calibration is performed on high-definition images, then image calibration accuracy is improved, but processing time increases significantly

Engineering Contradiction:
Improvecoordinate calibration accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The image is divided into multiple blocks, and coordinate calibration is performed on block vertices rather than all pixels. This segmentation approach reduces the number of calibration points from thousands of pixels to a manageable number of block corners, significantly decreasing processing time while maintaining calibration accuracy through subsequent interpolation of pixel coordinates within each block.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If pixel-by-pixel coordinate calibration is performed on high-definition images, then image calibration accuracy is improved, but memory capacity requirements increase

Engineering Contradiction:
Improvecoordinate calibration accuracyVSAvoidmemory capacity
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

By segmenting the image into blocks and performing calibration only on block vertices, the memory required to store calibration data is dramatically reduced. Instead of storing calibration information for every pixel in a high-definition image, the system stores coordinates only for block corners and uses interpolation formulas to calculate pixel coordinates on-demand, reducing memory requirements from megabytes to kilobytes.

Inventive Principle:
Principle #1Segmentation

3Adaptability or versatility

If the camera lens is disposed at any position for flexible installation, then adaptability is improved, but image tilt increases requiring calibration

Engineering Contradiction:
Improvelens position flexibilityVSAvoidcoordinate calibration complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The segmentation method simplifies the calibration process for arbitrarily positioned cameras by dividing the image into blocks and calibrating only block vertices. This approach maintains the flexibility of arbitrary lens positioning while reducing calibration complexity from pixel-by-pixel operations to block-level operations, making the system adaptable to various installation positions without proportionally increasing complexity.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS10255657B2Image perspective conversion method by converting coordinates of polygonal sub-regions and system thereof
Publication Date: 2019.04.09 WELTREND SEMICON INC
  • US10255657B2 patent drawing
  • US10255657B2 patent drawing
  • US10255657B2 patent drawing

AI summary

An image perspective conversion method includes acquiring a first image, partitioning a predetermined region on the first image into a plurality of polygonal sub-regions, acquiring a plurality of first coordinates corresponding to the plurality of polygonal sub-regions, converting the plurality of first coordinates into a plurality of second coordinates according to a world coordinate system, and interpolating a plurality of pixels among the plurality of second coordinates for generating a second image.