Title: Elimination of grey level distortion using multiscale gradient multiplication

Authors: B. Karthikeyan; Sneha Ballakur; S. Sowvarnica; V. Vaithiyanathan; N. Vinayakaram; G. Vasanth

Addresses: School of Computing, SASTRA University, Thanjavur-613401, India ' School of Computing, SASTRA University, Thanjavur-613401, India ' School of Computing, SASTRA University, Thanjavur-613401, India ' School of Computing, SASTRA University, Thanjavur-613401, India ' School of Computing, SASTRA University, Thanjavur-613401, India ' School of Computing, SASTRA University, Thanjavur-613401, India

Abstract: Pixels are the smallest solitary constituent of a digital image. Breaking down an image into its composite pixels elucidates its properties and subjecting it to thresholding helps to perceive the decomposition of its pixels rationally. This is done in assent with the pixel properties based on a threshold value. However, thresholding does not assure that these decomposed pixels are always pure. The odds of discerning the distortion (noise) enmeshed in the grey scale values are high. Hence removal of noise from the image is given paramount importance. Multiscale gradient multiplication is a method to obtain optimal threshold value which can be used to effectively eliminate grey level distortion upon segmentation. This method multiplies the responses from various filters operating at different scales. The proposed technique assists in eliminating distortion in grey levels while improving robustness towards edge signals, and thus, considerably outperforms other conventional methods.

Keywords: thresholding; multiscale gradient multiplication; transition region; filters; image segmentation; misclassification errors; grey level distortion; digital images; edge signals; image processing.

DOI: 10.1504/IJCVR.2016.079397

International Journal of Computational Vision and Robotics, 2016 Vol.6 No.4, pp.369 - 380

Received: 20 May 2014
Accepted: 15 Oct 2014
Published online: 28 Sep 2016 *

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