干旱区生态环境知识资源中心(Arid): Low dimensional multi-block neighborhood combination pattern for biomedical image retrieval

Arid

DOI	10.1007/s11042-022-12089-7
	Low dimensional multi-block neighborhood combination pattern for biomedical image retrieval
	Wadhera, Ankita; Agarwal, Megha
通讯作者	Agarwal, M
来源期刊	MULTIMEDIA TOOLS AND APPLICATIONS
ISSN	1380-7501
EISSN	1573-7721
出版年	2022
卷号	81 期号:19 页码:27853-27877
英文摘要	Content based image retrieval (CBIR) has been a thrust area of research to retrieve relevant images quickly from a huge image database. In this pursuit, a low dimensional multi-block neighborhood combination pattern (MNCP) is proposed for biomedical image retrieval. Traditional local binary pattern (LBP) failed to capture the macro-structures present in the image. A multi-block technique is applied here to design a feature insensitive to noise. Further, MNCP computes the modified Weber's ratio by encoding three different combinations of change in intensities among pixels to obtain unique patterns. This process considers sign and magnitude both of intensity changes and hence, the direction of intensity changes is also incorporated. In order to make the feature robust, these three combination patterns are concatenated. The most significant features of MNCP are selected to provide maximum inter class separability and variance using principal component analysis (PCA) and linear discriminant analysis (LDA) algorithms. Experiments are conducted on four very distinct and popular medical image datasets namely: OASIS MRI, VIA/I-ELCAP CT, Emphysema CT and MESSIDOR retinal database to examine the ability of the proposed method. Results of the proposed approach proves its superiority by outperforming the existing handcrafted as well as deep learning techniques in terms of average retrieval precision (ARP), average retrieval rate (ARR) and mean average precision (MAP). The proposed CBIR system takes very less time in retrieving the relevant images hence, it is suitable for real time applications as well.
英文关键词	Texture feature Biomedical image retrieval Local binary pattern Dimensionality reduction
类型	Article
语种	英语
收录类别	SCI-E
WOS记录号	WOS:000774644100013
WOS关键词	TERNARY COOCCURRENCE PATTERNS ; LINEAR DISCRIMINANT-ANALYSIS ; FEATURE DESCRIPTOR ; TEXTURE ; FACE ; EFFICIENT ; EIGENFACES ; SCALE ; MRI
WOS类目	Computer Science, Information Systems ; Computer Science, Software Engineering ; Computer Science, Theory & Methods ; Engineering, Electrical & Electronic
WOS研究方向	Computer Science ; Engineering
资源类型	期刊论文
条目标识符	http://119.78.100.177/qdio/handle/2XILL650/393826
推荐引用方式 GB/T 7714	Wadhera, Ankita,Agarwal, Megha. Low dimensional multi-block neighborhood combination pattern for biomedical image retrieval[J],2022,81(19):27853-27877.
APA	Wadhera, Ankita,&Agarwal, Megha.(2022).Low dimensional multi-block neighborhood combination pattern for biomedical image retrieval.MULTIMEDIA TOOLS AND APPLICATIONS,81(19),27853-27877.
MLA	Wadhera, Ankita,et al."Low dimensional multi-block neighborhood combination pattern for biomedical image retrieval".MULTIMEDIA TOOLS AND APPLICATIONS 81.19(2022):27853-27877.