Validating image-processing algorithms

Robert M. Haralick

doi:10.1117/12.387678

6 June 2000 Validating image-processing algorithms

Robert M. Haralick

Proceedings Volume 3979, Medical Imaging 2000: Image Processing; (2000) https://doi.org/10.1117/12.387678
Event: Medical Imaging 2000, 2000, San Diego, CA, United States

Abstract

This paper describes how to validate image processing algorithms when the algorithm has been designed to produce a numerical output that optimizes a given scalar criterion function having all first and second partial derivatives. The validation technique analyzes the statistical behavior of the algorithm under small random perturbations of its input. This paper gives the theory for the computation of the covariance matrix of the output when the input is perturbed by small random input perturbations. A statistical test is done examining the algorithm output under multiple instances of simulated perturbations and comparing these outputs with the ideal output and the analytically derived covariance matrix. A second level of simulations can be done that examines the test statistics of the hypothesis of the statistical test to determine whether in fact it is distributed in accordance with statistical theory. If the hypothesis is rejected that the output perturbations do not have the analytically predicted covariance matrix, then the image analysis software fails validation.

Citation Download Citation

Robert M. Haralick "Validating image-processing algorithms", Proc. SPIE 3979, Medical Imaging 2000: Image Processing, (6 June 2000); https://doi.org/10.1117/12.387678

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE

Includes PDF, HTML & Video, when available