.png){kind=icon}
.png){kind=icon}
Return
Content-based face image retrieval using quaternion based local diagonal extreme value pattern
.png){kind=icon}
PRE
AI
DOI：10.1007/s11042-023-15234-y.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper proposes a content based face image retrieval technique that utilizes Quaternion image representation and converts it into multiple images by using certain low and high pass frequency filters, afterwards a new descriptor named as local diagonal extreme value pattern (LDEVP) is computed for each filtered image and the obtained codes are finally combined in a single descriptor. LDEVP extracts the directional information from image by utilizing the indices and values of local diagonal extremes obtained through first order diagonal derivative. Usage of diagonal neighbors reduces the overall dimension of descriptor to a great extent. Simulations are performed on some standardized face datasets. The proposed technique attains superiority in terms of efficiency and performance by achieving precision of about 83.82%, 87.36%, 76.11% and 82.81% for AT&T, AR, Extended Yale and FERET datasets respectively.
Keywords:
Content based image retrieval
Quaternion
Face image
Local diagonal neighbors
Journal
IF:
3
Papers:
1.9W
Citations:
3.2W