Windana, Fredy (2014) Implementasi Kombinasi Feature Extraction untuk Content Based Image Retrieval. Magister thesis, Universitas Brawijaya.
Abstract
Kumpulan file citra pada suatu perangkat komputer dapat dicari berdasarkan metadata dari file citra seperti nama file , ukuran, kedalaman pixel , dan format ekstensi. Namun hasil menggunakan metadata kadang kala secara visualisasi tidak sesuai. Cara lain yang lebih spesifik berdasarkan ekstraksi ciri citra yakni memakai Content Based Image Retrieval . Metode pencarian dari CBIR dapat memakai metode tunggal seperti berdasarkan ekstraksi ciri warna atau tekstur saja maupun kombinasi antara ekstraksi ciri warna dan tesktur. Dalam penelitian ini aplikasi yang dibuat menggunakan metode kombinasi antara ekstraksi ciri LCH (HSV) dengan koefisien DC, kombinasi antara Statistik Orde 2 GLCM sub block dengan koefisien DC dan kombinasi antara LCH (HSV), Statistik Orde 2 GLCM sub block , dan koefisien DC. Citra sampel uji coba dari dataset CorelDB. Hasil uji coba pada penelitian ini menunjukkan nilai performa Non Interpolating Average Precision metode kombinasi antara LCH (HSV) dengan koefisien DC sekitar 23%. Sedangkan variasi kombinasi lain antara Statistik Orde 2 GLCM sub block dengan koefisien DC dan kombinasi antara LCH (HSV), Statistik Orde 2 GLCM sub block , dengan koefisien DC sekitar 12%.
English Abstract
An image collection on computer can be searched based on metadata of image file like file name, size, pixel depth, and format extension. However, results using metadata, sometimes in visualize not appropriate. Ano r way of more specific based on extraction description of image which is using Content Based Image Retrieval. CBIR method can used single method based on extraction description of color or texture of image or combination of extraction feature of color and texture. In this research, CBIR method used several combination of feature extraction methods, such as a combination between LCH (HSV) and DC Coefficient, a combination between 2 nd Order Statistic GLCM Sub Block and DC Coefficient, and a combination between LCH (HSV), 2 nd Order Statistic GLCM Sub Block and DC Coefficient. Sample image experiment test was used from CorelDB dataset. result from research showed performance component value of Non Interpolating Average Precision method performed by combination of LCH (HSV) and DC Coefficient performance rate was about 23%. While variance of combination of 2 nd order statistic GLCM sub block and coefficient DC and combination of LCH (HSV), 2 nd order statistic GLCM sub block and coefficient DC showed performance rate was about 12%.
| Item Type: | Thesis (Magister) |
|---|---|
| Identification Number: | TES/621.367/WIN/i/041407872 |
| Subjects: | 600 Technology (Applied sciences) > 621 Applied physics > 621.3 Electrical, magnetic, optical, communications, computer engineering; electronics, lighting |
| Divisions: | S2/S3 > Magister Teknik Elektro, Fakultas Teknik |
| Depositing User: | Endro Setyobudi |
| Date Deposited: | 12 Dec 2014 16:00 |
| Last Modified: | 12 Dec 2014 16:00 |
| URI: | http://repository.ub.ac.id/id/eprint/158659 |
Actions (login required)
 |
View Item |