Axis-Guided Vessel Segmentation Using a Self-Constructing Cascade-AdaBoost-SVM Classifier

Algorithm 1

The cascade-AdaBoost-SVM algorithm.

(1) For each layer, set the values for the minimum acceptable detection rate , the maximum acceptable false positive rate ,

and the maximum number of weak classifiers .

(2) Set the target overall false positive rate, .

(3) and are the positive and negative training sample sets.

(4) Initialize the cascade classifier: , , .

(5) While () or is not NULL

(i) , , , .

(ii) While () and and

(a) ;

(b) Use and to train an AdaBoost classifier with features.

(c) Evaluate current cascade classifier on and to calculate and .

(iii) If

(a) Use and to train a SVM classifier with features.

(b) Evaluate current cascade classifier on and to calculate and .

(iv) Remove the true negative detections from .