Segmentation of Endothelial Cell Boundaries of Rabbit Aortic Images Using a Machine Learning Approach

Table 4

Comparison between SVM and different conventional methods. Here the number of training pixels was 3844 with 2 classes, and the number of test pixels was 16384.

Method	Accuracy%
SVM with RBF kernel	94.09
Otsu [24]	82.22
Kittler [7]	82.21
Canny [22]	71.83