Thai Finger-Spelling Recognition Using a Cascaded Classifier Based on Histogram of Orientation Gradient Features

Pseudocode 1

The AdaBoost algorithm.

(i) Given images where for negative and positive examples.

(ii) Initialize weight , for respectively,

where and are the number of negatives and positive.

(iii) Set of weak classifier

(iv) Initial True Positive Rate and the False Positive Rate

For :

(1) Normalize the weights,

(2) the error of the weak classifier:

(3) Choose the classifier, , with lowest error

(4) Set and

    Update weights:

(v) The final strong classifier is: