Klasifikasi Otomatis Motif Tekstil Menggunakan Support Vector Machine Multi Kelas

Penulis

Ramadhani Ramadhani, Fitri Arnia, Rusdha Muharar

Abstrak

Tekstur merupakan pola atau motif tertentu yang tersusun secara berulang-ulang pada citra. Tekstur mudah dikenali/dikelompokkan oleh manusia, tetapi sulit bagi mesin. Klasifikasi tekstur secara otomatis berguna dan dibutuhkan pada banyak bidang seperti industri tekstil, pendaratan pesawat otomatis, fotografi dan seni. Pada industri tekstil, klasifikasi tekstur otomatis dapat meningkatkan efisiensi proses desain motif. Motif tekstil terdiri dari banyak kelompok, sehingga diperlukan metode klasifikasi multi kelas untuk mengelompokkan motif-motif tersebut. Artikel ini memaparkan kinerja tiga metode Support Vector Machine (SVM) multi kelas: One Against One (OAO), Directed Acyclic Graph (DAG) dan One Against All (OAA) pada klasifikasi motif dari citra tekstil, dimana Wavelet Gabor digunakan sebagai pengekstraksi fitur. Kinerja SVM diukur berdasarkan parameter akurasi dan fitur Gabor diekstraksi dengan skala dan orientasi yang berbeda. Tujuan penelitian ini adalah menentukan kinerja SVM dan pengaruh jumlah skala dan orientasi Gabor yang digunakan pada klasifikasi motif tekstil. Pada simulasi digunakan 120 citra tekstil yang terbagi menjadi tiga kategori motif: bunga, kotak dan polkadot. Akurasi pengelompokan SVM mencapai kisaran 90%-100%, bahkan untuk citra yang terpotong. Pengujian dengan k-fold validation menunjukkan bahwa SVM DAG lebih baik daripada SVM OAO dan SVM OAA, dengan akurasi mencapai 78%.

 

Abstract

Texture is a repetition of a specific pattern concatenation in an image. The Texture can be defined as a repetition of pattern in an image.  The texture is easy for the human to classify, but it is not easy for a machine. Automatic texture classification is useful and required in many fields such as textile industry, automatic aircraft landing, photography and art. In the textile industry, automatic texture classification can enhance the efficiency of motif designing process. The textile motif is various and should be grouped into more than two classes; therefore a multiclass classification is required. This article discusses the performance of multiclass Support Vector Machine (SVM): One Against One (OAO), Directed Acyclic Graph (DAG) and One Against All (OAA) in classifying textile motifs, in which the Gabor Filter was used to extract the texture features. The SVM performance was measured in terms of accuracy, while the Gabor features were extracted in a different combination of scales and orientations. The purpose of the work is to measure the SVM performance and determine the effect of using various Gabor scales and orientations in textile motifs classification. We used 120 textile images with three motifs: flower, boxes and polka dot. The SVM accuracy of 90%-100% was achieved; even for cropped textile images. Using the k-fold validation, the accuracy of SVM DAG was 78%, higher than those of SVM OAO and SVM OAA



Teks Lengkap:

PDF

Referensi


GHOSH, A. ET AL. (2011) ‘Pattern classification of fabric defects using support vector machines’, International Journal of Clothing Science and Technology. Emerald Group Publishing Limited, 23(2/3), pp. 142–151.

HOU, Z. AND PARKER, J. M. (2005) ‘Texture defect detection using support vector machines with adaptive gabor wavelet features’, in Application of Computer Vision, 2005. WACV/MOTIONS’05 Volume 1. Seventh IEEE Workshops on, pp. 275–280.

HOWARTH, P. AND RÜGER, S. M. (2004) ‘Evaluation of texture features for content-based image retrieval’, in CIVR, pp. 326–334.

HU, J. ET AL. (2012) ‘Fish species classification by color, texture and multi-class support vector machine using computer vision’, Computers and electronics in agriculture. Elsevier, 88, pp. 133–140.

JAIN, M. AND SINHA, A. (2015) ‘Classification of satellite images through gabor filter using svm’, International Journal of Computer Applications. Foundation of Computer Science, 116(7).

MESSIER, P. AND JOHNSON, C. R. (2014) ‘Automated surface texture classification of photographic print media’, in Signals, Systems and Computers, 2014 48th Asilomar Conference on, pp. 1105–1108.

MUCHTAR, M. AND CAHYANI, L. (2015) ‘Klasifikasi Citra Daun dengan Metode Gabor Co-Occurence’, ULTIMA Computing, 7(2), pp. 39--47.

NUGROHO, A. S., WITARTO, A. B. AND HANDOKO, D. (2003) ‘Support vector machine teori dan aplikasinya dalam bioinformatika’, Kuliah Umum Ilmu Komputer. com.

PAWENING, R. E. ET AL. (2015) ‘Classification of textile image using support vector machine with textural feature’, in Information & Communication Technology and Systems (ICTS), 2015 International Conference on, pp. 119–122.

PERMATA, E., PURNAMA, I. K. E. AND PURNOMO, M. H. (2013) ‘Klasifikasi Jenis dan Fase Parasit Malaria Plasmodium Falciparum dan Plasmodium Vivax dalam Sel Darah Merah Menggunakan Support Vector Machine One Against One’, SEMNASTEKNOMEDIA ONLINE, 1(1), pp. 1–2.

PRASETYO, E. (2014) ‘Data mining mengolah data menjadi informasi menggunakan matlab’, Yogyakarta: Andi Offset.

PUTRA, D. (2010) Pengolahan citra digital. Yogyakarta: Penerbit Andi.

BEN SALEM, Y. AND NASRI, S. (2009) ‘Automatic classification of woven fabrics using multi-class support vector machine’, Research Journal of Textile and Apparel. Emerald Group Publishing Limited, 13(2), pp. 28–36.

SINGH, S. M. AND HEMACHANDRAN, K. (2012) ‘Content-based image retrieval using color moment and Gabor texture feature’, IJCSI Int. J. Comput. Sci. Issues9, 9(5), pp. 299–309.

SONG, A. ET AL. (2014) ‘A Novel Texture Sensor for Fabric Texture Measurement and Classification.’, IEEE Trans. Instrumentation and Measurement, 63(7), pp. 1739–1747.

SUPIANTO, A. A. S. (2015) ‘Klasifikasi Citra Satelit Menggunakan Kombinasi Fitur Warna Dan Fitur Tekstur’, Jurnal Teknologi Informasi dan Ilmu Komputer, 2(2), pp. 102–109.

WISESTY, U. N. (2016) ‘Implementasi Gabor Wavelet dan Support Vector Machine pada Deteksi Polycystic Ovary (PCO) Berdasarkan Citra Ultrasonografi’, Indonesian Journal on Computing (Indo-JC), 1(2), pp. 67–82.

YANG, W. ET AL. (2011) ‘Fast recognition of foreign fibers in cotton lint using machine vision’, Mathematical and Computer Modelling. Elsevier, 54(3–4), pp. 877–882.