Scholarlink

<p>Demand of robotic services and surveillance systems require better accuracy in the performance that led the improvements in automatic object classification systems. However, classification of depth images makes this task more challenging. Therefore, this paper introduces a new enhanced bag-of-feature framework for feature representation in the classification of RGB-D images. The proposed method uses the logarithmic spiral henry gas solubility optimization algorithm and K-means clustering algorithm to find the optimal clusters of the visual words. Moreover, probability based fuzzy gaussian mixture model is applied for soft clustering of the visual words. The competence of the new logarithmic spiral henry gas solubility optimization has been tested over 41 benchmark problems of different modularity. The proposed classification model tested on the objects of the University of Washington RGB depth dataset. The dataset consists of 300 household objects of 51 categories. The proposed method achieves an average accuracy of 76% in object-based classification. Comparative analysis highlights the effectiveness of the proposed method for depth image classification.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhanced bag of features using logarithmic spiral HGSO and probability based fuzzy Gaussian mixture model for RGB-D object classification

  • 期刊论文

  • 发表日期：

    2025年6月28日

查看全文
  • Nand Kishor Yadav,
  • Mukesh Saraswat

摘要

Demand of robotic services and surveillance systems require better accuracy in the performance that led the improvements in automatic object classification systems. However, classification of depth images makes this task more challenging. Therefore, this paper introduces a new enhanced bag-of-feature framework for feature representation in the classification of RGB-D images. The proposed method uses the logarithmic spiral henry gas solubility optimization algorithm and K-means clustering algorithm to find the optimal clusters of the visual words. Moreover, probability based fuzzy gaussian mixture model is applied for soft clustering of the visual words. The competence of the new logarithmic spiral henry gas solubility optimization has been tested over 41 benchmark problems of different modularity. The proposed classification model tested on the objects of the University of Washington RGB depth dataset. The dataset consists of 300 household objects of 51 categories. The proposed method achieves an average accuracy of 76% in object-based classification. Comparative analysis highlights the effectiveness of the proposed method for depth image classification.