(For USM Staff/Student Only)

EngLib USM > Ω School of Electrical & Electronic Engineering >

Automatic number of clusters determination of clustering algorithms for image segmentation application

Automatic number of clusters determination of clustering algorithms for image segmentation application / Tang Jing Rui
Peruasan imej melibatkan pembahagian imej semula jadi kepada kawasan yang mempunyai tekstur homogen. Disebabkan oleh kemudahan pelaksanaan, kecekapan dan keupayaan untuk menjana penyelesaian yang memuaskan dalam banyak aplikasi, algoritma pengelompokan konvensional, iaitu Purata-K (K-Means) dan Purata-C Fuzi (Fuzzy C-Means) digunakan secara meluas dalam peruasan imej. Namun, algoritma pengelompokan tersebut sensitif terhadap pememulaan awal seperti penentuan bilangan kelompok, di mana bilangan kelompok perlu disediakan oleh pengguna algoritma. Dalam kajian ini, satu versi penambahbaikan bagi teknik peruasan imej berdasarkan analisa histogram dicadangkan. Empat teknik pelicinan histogram, iaitu teknik regresi setempat, teknik regresi setempat versi lasak, penapis laluan rendah dan penapis-licin Savitzky-Golay digunakan pada histogram imej untuk menentukan bilangan kelompok secara automatik. Kedua-dua analisa kualitatif dan kuantitatif membuktikan bahawa teknik pelicinan histogram berupaya untuk menyediakan bilangan kelompok yang optimum bagi algoritma pengelompokan konvensional. Keputusan ujikaji menunjukkan bahawa bagi algoritma pengelompokan Purata-K, imej keluaran peruasan yang terbaik dihasilkan apabila bilangan kelompok ditentukan oleh histogram yang dilicinkan dengan teknik regresi setempat versi lasak. Sebaliknya, bilangan kelompok yang optimum bagi algoritma pengelompokan Purata-C Fuzi dihasilkan apabila penapis laluan rendah digunakan untuk melicinkan histogram. Imej keluaran peruasan yang dihasilkan daripada algoritma yang dicadangkan berupaya meningkatkan kontras, mengekalkan jumlah maklumat terkandung yang tinggi dan menghasilkan kawasan ruasan yang lebih homogen. Penemuan tersebut menunjukkan bahawa integrasi teknik pelicinan histogram dengan algoritma pengelompokan konvensional untuk menentukan bilangan kelompok secara automatik dalam imej tanpa pengetahuan awalan pada imej berpotensi tinggi untuk kajian masa hadapan. _______________________________________________________________________________________________________ Image segmentation involves a process of partitioning a natural image into regions with homogeneous texture. Due to their ease of implementation, efficiency and capability in providing promising solution in many applications, the conventional clustering algorithms, K-Means and Fuzzy C-Means clustering algorithms are widely used in image segmentation. However, these clustering algorithms are sensitive to the initialization condition of number of clusters, where the number of clusters has to be provided by the user. In this study, an enhanced version of image segmentation technique based on histogram analysis is proposed. Four histogram smoothing techniques, namely local regression technique, robust version of local regression technique, low pass filter and Savitzky-Golay smoothing filter are applied on the histogram of the input image to determine the number of clusters automatically. Both qualitative and quantitative analyses prove that histogram smoothing techniques are capable of providing the optimum number of clusters for conventional clustering algorithms. Experimental results show that for K-Means clustering algorithm, the best segmented images are produced when the number of clusters is supplied by the histogram smoothed using the robust version of local regression technique. On the other hand, the optimum number of clusters for Fuzzy C-Means clustering algorithm is produced when low pass filter is used to smooth the histogram. The resultant segmented images produced by the proposed algorithms have successfully enhanced the contrast, preserved high amount of details and produced more homogeneous regions. These findings suggest that the integration of histogram smoothing techniques with the conventional clustering algorithms to automatically find the number of clusters in the image without prior knowledge on the image has high potential for future research.
Contributor(s):
Tang Jing Rui - Author
Primary Item Type:
Final Year Project
Identifiers:
Accession Number : 875004750
Language:
English
Subject Keywords:
Image; partitioning; homogeneous
First presented to the public:
6/1/2012
Original Publication Date:
7/2/2020
Previously Published By:
Universiti Sains Malaysia
Place Of Publication:
School of Electrical & Electronic Engineering
Citation:
Extents:
Number of Pages - 105
License Grantor / Date Granted:
  / ( View License )
Date Deposited
2020-07-02 16:20:05.575
Submitter:
Mohd Jasnizam Mohd Salleh

All Versions

Thumbnail Name Version Created Date
Automatic number of clusters determination of clustering algorithms for image segmentation application1 2020-07-02 16:20:05.575