Image Thresholding Berdasarkan Index Of Fuzziness Dan Fuzzy Similarity Measure

Pratamasunu, Gulpi Qorik Oktagalu (2015) Image Thresholding Berdasarkan Index Of Fuzziness Dan Fuzzy Similarity Measure. Masters thesis, Institut Teknologi Sepuluh Nopember.

[img]
Preview
Text
5113201021-Master_Thesis.pdf - Accepted Version

Download (6MB) | Preview

Abstract

Image thresholding merupakan proses yang penting pada beberapa aplikasi pemrosesan citra. Hasil thresholding yang buruk sering terjadi karena adanya derau dan pencahayaan yang tidak seragam pada citra. Beberapa metode telah diusulkan untuk mengatasi masalah ini dengan mengasumsikan citra sebagai fuzzy set. Namun, metode image thresholding yang penentuan threshold-nya hanya memperhatikan fungsi keanggotaan fuzzy set saja tidak selalu menghasilkan threshold yang optimal karena tidak memperhatikan bentuk histogram. Oleh karena itu, pada penelitian ini diusulkan metode image thresholding dengan penentuan threshold berdasarkan similaritas antar gray level menggunakan fuzzy similarity measure yang mempertimbangkan fungsi keanggotaan fuzzy set dan bentuk histogram. Pada penelitian ini, fuzzy region pada histogram ditentukan secara otomatis berdasarkan index of fuzziness terbesar pada setiap gray level. Kemudian histogram dibagi menjadi tiga daerah yaitu daerah gelap, fuzzy region, dan daerah terang. Setiap anggota fuzzy region diklasifikasikan kedalam dua daerah lainnya menggunakan fuzzy similarity measure. Uji coba dilakukan pada citra sintetis berderau dan citra natural. Performa thresholding dievaluasi menggunakan misclassification error. Rata-rata nilai misclassification error mencapai 6,20% pada citra sintetis berderau dan 6,85% pada citra natural. Hasil uji coba menunjukkan bahwa metode yang diusulkan tahan terhadap derau dan efektif digunakan pada citra berderau dan citra natural. ===================================================================================================== Image thresholding is an important task in several image processing applications. Bad thresholding results often occur because of the noise and nonuniform illumination on the image. Several methods have been proposed to overcome this problem by assuming the image as fuzzy sets. However, fuzzy image thresholding method based on fuzzy membership function cannot guarantee the optimal threshold selection, because it does not consider the shape of histogram. Therefore, in this study we propose a new method to threshold the image according to the similarity between gray levels using fuzzy similarity measure that considers fuzzy membership function and shape of histogram. In this study, the fuzzy region in the histogram is determined by calculating the largest index of fuzziness at each gray level. Then the histogram is divided into three regions, namely the dark region, fuzzy region, and the bright region. Each member of the fuzzy region is classified into the other two region using fuzzy similarity measure. Proposed method is evaluated using syntetic images with noises and natural images. The thresholding performance will be measured using misclassification error. Average value of misclassification error on syntetic images with noises and natural images are 6.20% and 6.85% respectively. Experiment results illustrate the robustness and effectiveness of the proposed method.

Item Type: Thesis (Masters)
Additional Information: RTIf 621.367 Pra i
Uncontrolled Keywords: Image Thresholding, Index of Fuzziness, Fuzzy Similarity Measure
Subjects: T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques
Divisions: Faculty of Information Technology > Informatics Engineering > 55101-(S2) Master Thesis
Depositing User: Yeni Anita Gonti
Date Deposited: 02 Apr 2020 13:54
Last Modified: 02 Apr 2020 13:54
URI: http://repository.its.ac.id/id/eprint/75665

Actions (login required)

View Item View Item