A Study on Military Image Classification with Support Vector Machine
碩士 === 國防大學國防管理學院 === 國防資訊研究所 === 97 === In this paper, the recent research work and relative technologies on content-based image retrieval (CBIR) are introduced. And the characteristic capture is the key step of the image classification in the CBIR algorithms. The color and the textural property ar...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2008
|
Online Access: | http://ndltd.ncl.edu.tw/handle/41925857921153702211 |
id |
ndltd-TW-097NDMC1654017 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-097NDMC16540172015-10-13T13:08:48Z http://ndltd.ncl.edu.tw/handle/41925857921153702211 A Study on Military Image Classification with Support Vector Machine 支援向量機用於軍事影像判釋之研究 Yu-Sun Chen 陳幼蓀 碩士 國防大學國防管理學院 國防資訊研究所 97 In this paper, the recent research work and relative technologies on content-based image retrieval (CBIR) are introduced. And the characteristic capture is the key step of the image classification in the CBIR algorithms. The color and the textural property are also often used in image visual features. By extracting the low-level features of images, three kind of resolutions, seven kind of colors and the texture characteristic values and their combinations respectively are adopted as the input features of the support vector machine for classification. After the experimental analysis, in color and in textural property combination, after the standardization, in the 16X24 and in the 32X48 resolution the characteristic combination classification effect is the best by the LAB+LH; in the 64X96 resolution the best effect is by the LAB+HL+LH. But after joining the parameter optimization, the LAB+HL+HH in 16X24, the LAB+HL+LH in 32X48 and the LAB+HL in 64X96 presented the best characteristic combination classification effect. Overall, the generalized analysis among various characteristics combinations is best by the 16X24 resolution's classified effect. Jie-Guan Tso 左杰官 2008 學位論文 ; thesis 80 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國防大學國防管理學院 === 國防資訊研究所 === 97 === In this paper, the recent research work and relative technologies on content-based image retrieval (CBIR) are introduced. And the characteristic capture is the key step of the image classification in the CBIR algorithms. The color and the textural property are also often used in image visual features. By extracting the low-level features of images, three kind of resolutions, seven kind of colors and the texture characteristic values and their combinations respectively are adopted as the input features of the support vector machine for classification.
After the experimental analysis, in color and in textural property combination, after the standardization, in the 16X24 and in the 32X48 resolution the characteristic combination classification effect is the best by the LAB+LH; in the 64X96 resolution the best effect is by the LAB+HL+LH. But after joining the parameter optimization, the LAB+HL+HH in 16X24, the LAB+HL+LH in 32X48 and the LAB+HL in 64X96 presented the best characteristic combination classification effect. Overall, the generalized analysis among various characteristics combinations is best by the 16X24 resolution's classified effect.
|
author2 |
Jie-Guan Tso |
author_facet |
Jie-Guan Tso Yu-Sun Chen 陳幼蓀 |
author |
Yu-Sun Chen 陳幼蓀 |
spellingShingle |
Yu-Sun Chen 陳幼蓀 A Study on Military Image Classification with Support Vector Machine |
author_sort |
Yu-Sun Chen |
title |
A Study on Military Image Classification with Support Vector Machine |
title_short |
A Study on Military Image Classification with Support Vector Machine |
title_full |
A Study on Military Image Classification with Support Vector Machine |
title_fullStr |
A Study on Military Image Classification with Support Vector Machine |
title_full_unstemmed |
A Study on Military Image Classification with Support Vector Machine |
title_sort |
study on military image classification with support vector machine |
publishDate |
2008 |
url |
http://ndltd.ncl.edu.tw/handle/41925857921153702211 |
work_keys_str_mv |
AT yusunchen astudyonmilitaryimageclassificationwithsupportvectormachine AT chényòusūn astudyonmilitaryimageclassificationwithsupportvectormachine AT yusunchen zhīyuánxiàngliàngjīyòngyújūnshìyǐngxiàngpànshìzhīyánjiū AT chényòusūn zhīyuánxiàngliàngjīyòngyújūnshìyǐngxiàngpànshìzhīyánjiū AT yusunchen studyonmilitaryimageclassificationwithsupportvectormachine AT chényòusūn studyonmilitaryimageclassificationwithsupportvectormachine |
_version_ |
1717733400372051968 |