Arabella, Imagine Clara (2017) Temu Kembali Citra Makanan Menggunakan Representasi Multi Texton Histogram. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
5113100004-Undergraduate-Theses.pdf - Published Version Download (4MB) | Preview |
Abstract
Temu kembali citra merupakan suatu metode pencarian citra dengan melakukan perbandingan antara citra query dengan citra yang terdapat dalam database berdasarkan informasi yang ada. Temu kembali citra untuk pengenalan makanan sangat dibutuhkan untuk pasien diet. Dalam tugas akhir ini diusulkan suatu metode temu kembali citra makanan berdasarkan input berupa citra makanan yang dibandingkan dengan database citra makanan yang ada.
Tugas akhir ini secara khusus membahas mengenai perancangan sebuah sistem temu kembali citra makanan dengan representasi Multi-texton Histogram (MTH). Proses pertama dilakukan deteksi orientasi tekstur menggunakan metode Sobel Edge Detection. Setelah itu dilakukan kuantisasi warna pada ruang warna RGB. Serta deteksi Texton untuk tahap ekstraksi fiturnya. Untuk mendapatkan kemiripan citra, dihitung jarak antar citra dengan menggunakan distance metric. Setelah didapatkan jarak antar citra, diurutkan dari yang terdekat sampai yang terjauh jarak citranya untuk menentukan temu kembali citra. Hasil yang didapat adalah berupa ditemukannya citra-citra yang mirip dengan citra query.
Berdasarkan uji coba yang dilakukan pada citra, hasil pencarian citra mirip dengan rata-rata nilai precision terbaik sebesar 40,50% dan recall terbaik sebesar 8,61% pada 18 level orientasi dan 64 level kuantisasi warna.
======================================================================
Image retrieval is an image search method by performing a comparison between the query image and the image contained in the database based on the existing information. Image retrieval for feeding recognition is essential for dietary patients. In this final project proposed a method of retrieval of food image based on input in the form of food image compared with existing food image database.
This final project specifically discusses the design of a food image retrieval system with a Multi-texton Histogram (MTH) representation. The first step is detection of texture orientation using Sobel Edge Detection method. After that is done the color quantization on RGB color space. As well as Texton Detection for its feature extraction stage. To get the image resemblance, calculated the distance between the image using distance metric. Having obtained the distance between images, sorted from the nearest to the farthest distance of its image to determine image retrieval. The results obtained are in the form of finding images similar to the image of the query.
Based on the experiments performed on the image, the best precision of image retrieval were obtained with average precision 40,5% and recall 8,61% at 18 level orientation and 64 level quantization.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSIf Ara t |
| Uncontrolled Keywords: | Citra Makanan, Edge Detection, Multi Texton Histogram, Temu Kembali Citra, Food Image, Image Retrieval |
| Subjects: | T Technology > T Technology (General) Z Bibliography. Library Science. Information Resources > ZA Information resources > Z699.5 Information storage and retrieval systems |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Imagine Clara Arabella |
| Date Deposited: | 23 Aug 2017 01:33 |
| Last Modified: | 23 Aug 2017 01:33 |
| URI: | http://repository.its.ac.id/id/eprint/42441 |
Actions (login required)
 |
View Item |