Title: Research on the computer vision cracked eggs detecting method

Authors: Yeqin Wang

Addresses: Faculty of Electronic and Electrical Engineering, Huaiyin Institute of Technology, Huai'an, Jiangsu 223001, China

Abstract: To improve the accuracy of egg damage detection and the objectivity of characteristic parameter selection, the egg image grey cooccurrence matrix is constructed, a parameter system is composed of angular second-moment and contrast and variance and sum of variance to describe the egg crack. The egg image is reconstructed and decomposed with two-scale wavelet transform, a new 24-dimension characteristic parameter system is constructed by picking up the grey level cooccurrence matrix (GLCM) parameters of the reconstructed high-frequency subimage. The reduced dimensionality parameter system constituted by principal component is obtained by principal component analysis (PCA), egg damage detection contrast experiment of the three parameter system above is carried out, the experiment results show that the accuracy equalled to 96.67% based on the parameter system constituted by principal component, the characteristic parameter system based on multi-resolution can increase the accuracy of egg crack detection.

Keywords: cracked eggs detection; computer vision; multi-resolution; egg damage; egg images; grey level cooccurrence matrix; GLCM; wavelet transform; principal component analysis; PCA.

DOI: 10.1504/IJCAT.2014.066730

International Journal of Computer Applications in Technology, 2014 Vol.50 No.3/4, pp.215 - 219

Published online: 07 Feb 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article