Retina Blood Vessels Extraction With Parallel Method
碩士 === 國立彰化師範大學 === 電子工程學系 === 105 === The two way of blood vessels extraction will be compared by using the rate of accuracy in the thesis. One is only for thick blood vessels, and another is for thick-and-fine blood vessels. The thick blood vessels are easy one and fine blood vessels are the hard...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2017
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/e5y894 |
| Summary: | 碩士 === 國立彰化師範大學 === 電子工程學系 === 105 === The two way of blood vessels extraction will be compared by using the rate of accuracy in the thesis. One is only for thick blood vessels, and another is for thick-and-fine blood vessels. The thick blood vessels are easy one and fine blood vessels are the hard one for the vessels extraction algorithm. First we extract thick blood vessels and calculate accuracy. Then conduct the thick-and-fine blood vessels extraction algorithm and accuracy bases on the thick blood vessels extraction algorithm. At least, we compare the difference in accuracy and picture between the two ways for conclusion.
We take the thick blood vessels extraction algorithm to compare the difference between the execution time that speeds up by using parallel process technology and normal version by using series process technology.
In this thesis, we use DRIVE[8] as testbench, Intel core i7 3770 CPU for series process and NVIDIA Geforce GTX760 GPU for parallel process. The result shows that GPU is 9.12 times faster than CPU. The thick-and-fine blood vessels extraction is better on accuracy of blood vessels pixels but worse on accuracy and accuracy of none blood vessels pixels.
|
|---|