Novel application of neutrosophic logic in classifiers evaluated under region-based image categorization system.
详细信息
- 作者：Ju ; Wen.
- 学历：Doctor
- 年：2011
- 导师：Cheng,Heng-Da,eadvisorDyreson,Curtisecommittee memberAllan,Vicki H.ecommittee memberAllan,Stephen J.ecommittee memberChen,Yangquanecommittee member
- 毕业院校：Utah State University
- Department：Computer Science.
- ISBN：9781124567860
- CBH：3449483
- Country：USA
- 语种：English
- FileSize：14085236
- Pages：70
文摘
Neutrosophic logic is a relatively new logic that is a generalization of fuzzy logic. In this dissertation,for the first time,neutrosophic logic is applied to the field of classifiers where a support vector machine (SVM) is adopted as the example to validate the feasibility and effectiveness of neutrosophic logic. The proposed neutrosophic set is integrated into a reformulated SVM,and the performance of the achieved classifier N-SVM is evaluated under an image categorization system. Image categorization is an important yet challenging research topic in computer vision. In this dissertation,images are first segmented by a hierarchical two-stage self-organizing map (HSOM),using color and texture features. A novel approach is proposed to select the training samples of HSOM based on homogeneity properties. A diverse density support vector machine (DD-SVM) framework that extends the multiple-instance learning (MIL) technique is then applied to the image categorization problem by viewing an image as a bag of instances corresponding to the regions obtained from the image segmentation. Using the instance prototype,every bag is mapped to a point in the new bag space,and the categorization is transformed to a classification problem. Then,the proposed N-SVM based on the neutrosophic set is used as the classifier in the new bag space. N-SVM treats samples differently according to the weighting function,and it helps reduce the effects of outliers. Experimental results on a COREL dataset of 1000 general-purpose images and a Caltech 101 dataset of 9000 images demonstrate the validity and effectiveness of the proposed method.