Rahmahusna, Alfia (2019) Dynamic Particle Swarm Optimization dan K-means Clustering untuk Segmentasi Citra. Sarjana thesis, Universitas Brawijaya.
Abstract
Pada skripsi ini dibahas segmentasi citra dalam ruang warna HSV menggunakan algoritma gabungan Dynamic Particle Swarm Optimization dan K-means clustering (DPSOK) yang didasari oleh algoritma Dynamic Particle Swarm Optimization (DPSO) dan K-means clustering. Algoritma DPSOK terinspirasi oleh algoritma gabungan Particle Swarm Optimization dan K-means clustering (PSOK). Algoritma PSO mampu mengatasi kelemahan algoritma K-means clustering yang sangat sensitif terhadap pusat klaster awal dan cepat mengalami konvergen ke optimum lokal. Namun algoritma PSO membutuhkan waktu yang lebih panjang untuk konvergen ke solusi optimum global. Algoritma Dynamic Particle Swarm Optimization (DPSO) menggunakan bobot inersia dinamis dan parameter sosial dinamis untuk menghitung kecepatan partikel. Dengan menggunakan bobot inersia dinamis dan parameter sosial dinamis, DPSOK mampu mencari solusi optimum lokal dan global secara seimbang dengan waktu yang lebih singkat. Hasil percobaan menunjukkan bahwa algoritma DPSOK mampu mencapai waktu segmentasi yang lebih singkat dibandingkan algoritma PSOK dan hasil segmentasi citra yang lebih baik dibandingkan algoritma K-means clustering.
English Abstract
This final project discusses image segmentation in HSV color space by using the Dynamic Particle Swarm Optimization and K-means clustering (DPSOK) algorithm, which is combination of Dynamic Particle Swarm Optimization (DPSO) and K-means clustering. DPSOK algorithm is inspired by the ideas of Particle Swarm Optimization and K-means clustering (PSOK) algorithm. The PSO algorithm is able to overcome the disadvantages of K-means clustering, which is highly sensitive to the initial cluster centers and is easy to fall into local optima and miss the global optimum. However, PSO algorithm needs a longer time to converge towards the global optimum. The DPSO algorithm applies dynamical inertia weight and dynamical learning factors to determine particle velocity. By using dynamical inertia weight and dynamical learning factors, DPSOK can find both local as well as global optima in a shorter time. Experimental results show that DPSOK algorithm needs a shorter computation time as compared to PSOK algorithm and gives better image segmentation results than K-means clustering algorithm.
Other obstract
-
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/MIPA/2019/10/051910713 |
| Uncontrolled Keywords: | segmentasi citra, ruang warna HSV, Particle Swarm Optimization (PSO), K-means clustering, PSOK. image segmentation, HSV color space, Particle Swarm Optimization (PSO), K-means clustering. |
| Subjects: | 500 Natural sciences and mathematics > 518 Numerical analysis > 518.1 Algorithms |
| Divisions: | Fakultas Matematika dan Ilmu Pengetahuan Alam > Matematika |
| Depositing User: | Budi Wahyono Wahyono |
| Date Deposited: | 24 Aug 2020 07:33 |
| Last Modified: | 28 Mar 2022 02:46 |
| URI: | http://repository.ub.ac.id/id/eprint/176647 |
Preview |
Text
Alfia Rahmahusna.pdf Download (2MB) | Preview |
Actions (login required)
 |
View Item |