Adaptif Range-Constrained Otsu Untuk Pemilihan Threshold Secara Otomatis Pada Histogram Citra Dengan Variansi Kelas Yang Tidak Seimbang

Authors

  • Gama Wisnu Fajarianto Magister Teknik Informatika, Fakultas Teknologi Informasi, ITS, Surabaya
  • Ahmad Hifdhul Abror Sistem Informasi, Fakultas Saintek, UIN Sunan Ampel, Surabaya
  • Nur Hayatin Teknik Informatika, Fakultas Teknik, Universitas Muhammadiyah Malang

DOI:

https://doi.org/10.26594/register.v2i1.439

Abstract

Abstrak Image Thresholding merupakan proses segmentasi untuk pemisahkan foreground dan background pada citra dengan cara membagi histogram citra menjadi dua kelas. Beberapa metode thresholding seperti Otsu dan Range-constrained Otsu menggunakan nilai variansi dari histogram untuk mendapatkan titik threshold, namun ketika menangani citra yang memiliki nilai variansi kelas foreground dan background tidak seimbang titik threshold yang dihasilkan kurang tepat. Paper ini mengusulkan metode Adaptif Range-constrained Otsu untuk mengatasi permasalahan variansi kelas yang tidak seimbang dengan cara mencari kelas yang memiliki nilai variansi lebih besar, untuk mendapatkan titik threshold yang lebih tepat. Pengujian menggunakan 22 NDT image dengan evaluasi misclassification error rate dan metode perankingan menunjukkan metode ini menghasilkan rerata ME 0.1153. Sedangkan Otsu sebesar 0.1746. Nilai rerata ranking 3.55, selisih 0.05 dibanding Kittler III. Hasil ini menunjukkan metode yang diusulkan kompetitif, terutama untuk segmentasi citra yang memiliki variansi kelas tidak sama. Kata kunci: segmentasi, thresholding, histogram, Otsu, Range-constrained. Abstract Image thresholding is segmentation process for separating foreground and background of an image by dividing image histogram into two classes. Several thresholding methods like Otsu and Rangeconstrained Otsu using the variance value of the histogram to get the threshold point, but when handling images that have unbalance class variance of the foreground and background produce less accurate threshold point. This paper proposes a method Adaptive Range-constrained Otsu to solve unbalance class variance problem by finding a class that has greater variance value to obtain more accurate threshold point. NDT testing using 22 images with misclassification error rate evaluation and ranking methods shows that this method results ME average of 0.1153, while Otsu method results 0.1746. The rankings mean value is 3.55, which has the difference of 0.05 when compared with Kittler III. These results show that the proposed method is competitive, especially for image segmentation with different class variance. Key word: segmentasi, thresholding, histogram, Otsu, Range-constrained.

References

Beauchemin, M. (2013). Image Thresholding Based on Semivariance. Pattern Recognition Letters, 34(5), 456??462.

Fan, J. L., & Lei, B. (2012). A Modified Valley-Emphasis Method for Automatic Thresholding. Pattern Recognition Letters, 33(6), 703??708.

Gao, X., Fu, R., Li, X., Tao, D., Zhang, B., & Yang, H. (2011). Aurora Image Segmentation by Combining Patch and Texture Thresholding. Computer Vision and Image Understanding, 115(3), 390??402.

Hou, Z., Hu, Q., & Nowinski, W. (1732??1743). On Minimum Variance Thresholding. Pattern Recognition Letters, 27(14), 2006.

Liu, C. C., Tsai, C. Y., Liu, J., Yu, C. Y., & Yu, S. S. (2012). A Pectoral Muscle Segmentation Algorithm for Digital Mammograms Using Otsu Thresholding and Multiple Regression Analysis. Computers & Mathematics with Applications, 64(5), 1100??1107.

Otsu, N. (1979). Thresholds Selection Method form Grey-Level Histograms. IEEE Trans. On Systems, Man and Cybernetics, 9(1), 62-66.

Qiao, Y., Hu, Q., Qian, G., Luo, S., & Nowinski, W. L. (2007). Thresholding Based on Variance and Intensity Contrast. Pattern Recognition, 40(2), 596??608.

Xu, X., Xu, S., Jin, L., & Song, E. (2011). Characteristic Analysis of Otsu Threshold and Its Applications. Pattern Recognition Letters, 32(2011), 956-961.

Yao, H., Duan, Q., Li, D., & Wang, J. (2013). An improved K-means Clustering Algorithm for Fish Image Segmentation. Mathematical and Computer Modelling, 58(3-4), 790??798.

Published

2016-01-01

How to Cite

[1]
G. W. Fajarianto, A. H. Abror, and N. Hayatin, “Adaptif Range-Constrained Otsu Untuk Pemilihan Threshold Secara Otomatis Pada Histogram Citra Dengan Variansi Kelas Yang Tidak Seimbang”, Register: Jurnal Ilmiah Teknologi Sistem Informasi, vol. 2, no. 1, pp. 6–10, Jan. 2016.

Issue

Section

Article