Cosine modulated filter bank‐based architecture for extracting and fusing saliency features

Author:

Ali Md. Yousuf1,Jiang Bin1,Chowdhury Oindrila2,Harun‐Ar‐Rashid Md.3,Hossain M. Shamim4ORCID,AlMutib Khalid4

Affiliation:

1. Department of Computer Engineering, College of Computer Science and Electronic Engineering Hunan University Changsha China

2. Department of Computer Science and Engineering American International University‐Bangladesh (AIUB) Dhaka Bangladesh

3. Department of Computer Science and Engineering Mawlana Bhashani Science and Technology University Tangail Bangladesh

4. Department of Software Engineering, College of Computer and Information Sciences King Saud University Riyadh Saudi Arabia

Abstract

AbstractMany academics are interested in content‐based image retrieval techniques like image segmentation. In computer vision, the most popular method for segmenting a digital image into different parts is known as image segmentation. We assigned the artificially intelligent algorithm to the image's critical areas by modeling human features in specific regions. In order to detect the object and identify the key parts in the ‘RGB’ photographs, we combined scenes based on a colour and depth map, or ‘RGB‐D’, and used cosine modulated filter bank (CMFB), which conducts cross‐scale extraction of joint features from the images during feature extraction. The proposed ‘CMFB’ combines the discovered collaborative elements with the discovered supplementary data. The features in multi‐scale images is combined using fusion blocks with the goal of producing additional features (FB). Then, a saliency mapping calculation is made for the loss linked to two blocks. The suggested ‘CMFB’ is tested with the aid of five data sets, and it is shown that, the proposed ‘CMFB’ outperforms other conventional techniques.

Publisher

Wiley

Subject

Artificial Intelligence,Computational Theory and Mathematics,Theoretical Computer Science,Control and Systems Engineering

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3