Feature Classification in Ultrasound Textures for Image Quality Assessment:a Preliminary Study on the Characterization and Selection of Haralick Parameters by Means of Correlation Matrices

L. Schinaia, A. Scorza , F. Orsini,S. A. Sciuto

Abstract

This paper describes a preliminary study on feature selection from the gray level co-occurrence matrix (GLCM) among the 14 features proposed by R.M. Haralick (1979) with the aim to apply them to ultrasound image classification and Quality Assessment. In particular4 main-classes of images with different patterns (Lines, Chess, alternates Row and Circles)have been implemented and different levels ofspeckle noisehave been added to simulate ultrasound images with different textures.With the aim to characterize therelationship betweenHaralickfeatures and the pattern type, size, contrastand noise, someCorrelation Matrices have been implemented. Preliminary results are explained and discussed.

Keywords

diagnostic ultrasound, Haralick textural features, uniformity, B-mode, quality assessment

Download

IMEKO-TC4-2017-032.pdf

DOI

-

IMEKO TC

TC4 - Measurement of Electrical Quantities

Back to the proceedings