Automic Edge Detection in Noisy Fringe Patterns: A Practical Approach

Authors

  • Mani Maran Ratnam

DOI:

https://doi.org/10.11113/jt.v37.541

Abstract

Corak-corak pinggir biasanya dikaji dengan menggunakan tingkap yang ditakrifkan awal kerana kesukaran mencari sempadannya. Hal ini berlaku kerana pinggir-pinggir tersebut biasanya mempunyai keamatan minimum yang serupa dengan keamatan latar belakang dan adalah sukar untuk menentukan di mana sesuatu pinggir bermula dan berakhir. Kehadiran pinggir-pinggir boleh menyebabkan ralat dalam pengesanan sempadan dengan menggunakan kaedah-kaedah pengesanan sempadan yang sedia ada. Kertas kerja ini membentangkan suatu pendekatan praktik ke arah pengesanan automatik sempadan corak pinggir berhingar dengan paras hingar yang berbeza. Teknik tersebut menggunakan operasi-operasi morfologi pengembangan dan penghakisan serta pengesan sempadan binari kejiranan mudah. Nilai ambang yang diperlukan dalam operasi binarisasi ditentukan daripada hubungkait linear antara nilai ambang dengan sisihan piawai hingar Gaussian daripada kajian simulasi. Teknik tersebut kemudian dilaksanakan dengan jayanya pada corak-corak pinggir pada beberapa objek sebenar yang didapati melalui kaedah pengunjuran pinggir. Beberapa kelemahan kaedah tersebut dibincangkan. Kata kunci: Pengesanan sempadan, corak pinggir Fringe patterns are usually analyzed using a predefined window due to the difficulty associated with finding their edges. This is because, the fringes usually have minimum intensities similar to that of the background and it is difficult to determine where a fringe starts or ends. The presence of the fringes can therefore lead to errors in detecting the edges of the object when using existing edge detection techniques. This paper presents a practical approach towards the automatic detection of the edges of a noisy fringe pattern having different noise levels. The technique uses the dilation and erosion morphological operators combined with a simple neighborhood binary edge detector. The threshold required for the binarization operation is determined from a linear relationship between the threshold value and the standard deviation of Gaussian noise from a simulation study. The proposed technique has been successfully applied to fringe patterns on real objects obtained using the fringe projection method. Some of the limitations of the technique are discussed. Key words: Edge detection; fringe patterns

Downloads

Published

2012-01-20

Issue

Section

Science and Engineering

How to Cite

Automic Edge Detection in Noisy Fringe Patterns: A Practical Approach. (2012). Jurnal Teknologi (Sciences & Engineering), 37(1), 45–58. https://doi.org/10.11113/jt.v37.541