Research Article
Blood Vessel Extraction in Color Retinal Fundus Images with Enhancement Filtering and Unsupervised Classification
Table 1
The effect of different cluster numbers to the performance results using hard and soft clustering after Gabor filter.
| | TPR | SPE | ACC |
| STARE images | Gabor + FCM with 2 clusters | 51.59 | 99.34 | 95.74 | Gabor + FCM with 3 clusters | 75.38 | 96.83 | 95.18 | Gabor + FCM with 4 clusters | 87.06 | 89.79 | 89.53 | Gabor + K-means with 2 clusters | 50.37 | 99.39 | 95.70 | Gabor + K-means with 3 clusters | 68.69 | 98.16 | 95.94 | Gabor + K-means with 4 clusters | 84.30 | 92.05 | 91.44 | DRIVE images | Gabor + FCM with 2 clusters | 54.60 | 99.01 | 93.34 | Gabor + FCM with 3 clusters | 67.76 | 98.59 | 95.87 | Gabor + FCM with 4 clusters | 54.50 | 99.02 | 93.34 | Gabor + K-means with 2 clusters | 54.51 | 99.01 | 93.33 | Gabor + K-means with 3 clusters | 61.02 | 99.05 | 95.71 | Gabor + K-means with 4 clusters | 54.44 | 99.02 | 93.33 |
|
|