Ellipsoid support vector machines for rotational texture classification
碩士 === 朝陽科技大學 === 資訊工程系碩士班 === 96 === In this paper we propose a modified framework for support vector machines, called Ellipsoid Support Vector Machines (ESVMs), to improve classification capability. The principle of ESVMs is to use a minimum ellipsoid to enclose the specific patterns. With this me...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2008
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/81707630229946107921 |
| Summary: | 碩士 === 朝陽科技大學 === 資訊工程系碩士班 === 96 === In this paper we propose a modified framework for support vector machines, called Ellipsoid Support Vector Machines (ESVMs), to improve classification capability. The principle of ESVMs is to use a minimum ellipsoid to enclose the specific patterns. With this method maximizing the margin of separation and minimizing the volume of ellipsoid are formulated as the regularized risk function. By adopting an efficient algorithm the proposed algorithm in this paper can be used with nonlinear kernels and has a time complexity that is O(N). Experimental results demonstrated that the ESVMs have comparable performance with existing SVM models.
In addition, a rotational texture classification method based on multi-model feature integration by ESVMs is proposed. Three feature sets based on three texture models––the Gabor filter, neighboring gray level dependence matrix (NGLDM) and minimum volume enclosing ellipsoids are used to describe the image properties. To classify image, combining multiple ESVMs classifiers using the intensity of decision function value is presented. Twenty five Brodatz textures were used to evaluate the classification performance. Experimental results demonstrated that high performance can be achieved by our proposed system.
|
|---|