Pre-processing Techniques for Detection of Blurred Images

被引：8

作者：

Francis, Leena Mary ^{[1
]}

Sreenath, N. ^{[1
]}

机构：

[1] Pondicherry Engn Coll, Dept Comp Sci & Engn, Pondicherry, India

来源：

PROCEEDINGS OF INTERNATIONAL CONFERENCE ON COMPUTATIONAL INTELLIGENCE AND DATA ENGINEERING (ICCIDE 2018) | 2019年 / 28卷

关键词：

Blur detection; Blur estimation; Gaussian Filter; Laplacian function; Threshold fixing;

D O I：

10.1007/978-981-13-6459-4_7

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Blur detection and estimation have progressively became an imminent arena of computer vision. Along with heightening usage of mobiles and photographs, detecting the blur is purposed over to enhance or to remove the images. PrE-processing Techniques for DEtection of Blurred Images(PET-DEBI) was framed to detect the blurred and undistorted images. The frailty of Laplacian has been overcome by Gaussian filter to remove the noise of the image; then, the variance of Laplacian is calculated over the images. Through analysing the variance of the images, appropriate threshold is circumscribed and further used as limitation to define blurred and unblurred images. PET-DEBI was implemented and experimented yielding encouraging results with accuracy of 87.57%, precision of 88.88%, recall of 86.96% and F-measure of 87.91%.

引用

页码：59 / 66

页数：8