Journal
MULTIMEDIA TOOLS AND APPLICATIONS
Volume 78, Issue 24, Pages 34839-34865Publisher
SPRINGER
DOI: 10.1007/s11042-019-08111-0
Keywords
Blood vessel segmentation; Clifford algebra; Clifford convolution; Fundus images; Multi-vector; RGB color model
Ask authors/readers for more resources
The appearance and structure of blood vessels in retinal fundus image is a fundamental part of diagnosing different issues related with such as diabetes and hypertension. The proposed blood vessel segmentation in fundus image using Clifford Algebra approach is divided into three steps. Image vectorization as a first step helps to convert the image space into Clifford space. Next step introduces Clifford matched filter as a proposed mask which works for retinal blood vessel extraction. The third and final step of this method is Clifford convolution operation with the help of Clifford convolution. This mask generates edge points along the boundaries of the blood vessels. The edge points are represented as a Grade-0 vector or scalar unit. Discrete edge points along the boundary of blood vessels are the edge pixels instead of continuous edges. The output of this method differs in the representation of vessel tree compare to other existing methods. The output image can be defined as the edge point set. This method achieves blood vessel segmentation accuracy of 94.88% and 92.95% on two publicly available datasets STARE and DRIVE respectively in less than 0.5 s per image. The proposed matched filter and the segmentation technique opens many windows of reliable and faster processing for further image processing steps on retinal fundus images.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available