Related Experiment Video
Updated: May 29, 2026
07:45
Quantifying Intermembrane Distances with Serial Image Dilations
Published on: September 28, 2018
Edge location to subpixel values in digital imagery.
1School of Electrical Engineering, Purdue University, West Lafayette, IN 49707; Bell Laboratories, Holmdel, NJ 07733.
IEEE Transactions on Pattern Analysis and Machine Intelligence
|August 27, 2011
Summary
A novel edge detection method precisely locates edges in digital data to subpixel values, remaining invariant to data changes. This robust technique accurately identifies edges even with additive noise, outperforming existing operators.
Area of Science:
- Digital Image Processing
- Computer Vision
- Computational Geometry
Background:
- Accurate edge detection is crucial for image analysis.
- Existing methods struggle with noise and subpixel precision.
- Invariance to illumination changes is a key challenge.
Purpose of the Study:
- Introduce a new subpixel edge detection method.
- Ensure invariance to additive and multiplicative data changes.
- Evaluate performance against established edge detection operators.
Main Methods:
- Fitting an ideal edge to data by matching moments for 1D patterns.
- Relating edge location to Christoffel numbers.
- Extending the method to 2D patterns using line equations.
- Analyzing the impact of additive noise on edge localization.
Main Results:
- The new method achieves subpixel edge localization.
- Demonstrated invariance to additive and multiplicative data variations.
- The 2D method shows comparable or superior performance to Hueckel operator.
- Performance evaluation against Sobel and Hueckel operators in noisy conditions.
Conclusions:
- The presented method offers robust and precise edge detection.
- It provides significant advantages in handling noise and illumination variations.
- This technique is a valuable advancement for digital image analysis applications.