Hough Transform Circle Detection Reducing Random Memory Access

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The Hough transform for circle detection in images requires significant execution time and memory, leading to inefficient processing and frequent random accesses, especially in embedded systems, and existing modifications do not adequately address memory requirements or performance when dealing with concentric circles or arcs.

Innovation Solution

A two-stage approach is adopted, where center candidates are first determined by searching the image line by line, and then the radius is estimated using a one-dimensional accumulator array that fits in local memory, reducing random accesses and computations through a filtering step that eliminates unnecessary lines and a voting step that accumulates votes line by line.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a three-dimensional accumulator array is used for Hough transform circle detection, then detection accuracy is improved, but memory requirements and execution time increase significantly

Engineering Contradiction:
Improvecircle detection accuracyVSAvoidmemory requirements
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent divides the circle detection process into two separate stages: first detecting line segments using a 2D accumulator array, then detecting circles using a 1D accumulator array. This segmentation reduces the dimensionality of the accumulator arrays from 3D to 1D and 2D respectively, significantly reducing memory requirements while maintaining detection accuracy through the two-stage approach

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from a 3D accumulator array (x, y, r) to a 2D accumulator array for line detection (x, θ) and then to a 1D accumulator array for circle detection (r). This dimensionality reduction is achieved by first detecting lines and then using those line parameters to constrain the circle search space, thereby reducing memory requirements

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If a three-dimensional accumulator array is used for Hough transform circle detection, then detection accuracy is improved, but processing time increases

Engineering Contradiction:
Improvecircle detection accuracyVSAvoidexecution time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the circle detection into two stages: line detection followed by circle detection. This segmentation allows each stage to use optimized accumulator arrays (2D for lines, 1D for circles) rather than a full 3D array, reducing the computational complexity and execution time while maintaining accuracy

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary line detection before circle detection. By first identifying line segments and their parameters (x, θ), the algorithm constrains the subsequent circle search space, eliminating the need to search all possible circles in a 3D parameter space and thereby reducing processing time

Inventive Principle:
Principle #10Preliminary action

3Ease of operation

If frequent random accesses to the accumulator array are performed, then circle detection can be implemented, but performance on embedded systems deteriorates

Engineering Contradiction:
Improvedetection functionalityVSAvoidprocessing efficiency
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

The patent reduces the accumulator array from 3D to 2D and then to 1D, which significantly reduces the random access memory requirements. The 1D accumulator array for circle detection can be implemented in cache memory or even registers on embedded systems, eliminating frequent external memory accesses and improving processing efficiency

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentEP2738711B1Hough transform for circles
Publication Date: 2023.12.27 ANALOG DEVICES INC
  • EP2738711B1 patent drawingFigure 1
  • EP2738711B1 patent drawingFigure 2~3
  • EP2738711B1 patent drawingFigure 4A~5

AI summary

The Hough transform for circles can be implemented in a manner that avoids random access to the Hough accumulator array by successively identifying center candidates in each line of the image based on edge pixels in corresponding lines voting on the line of interest.