Ensemble Evaluation for Image Segmentation using a Semi-Supervised SVM

Sang Jun Lee, Sang-Gyu Ryu, Yong-Ju Jeon, Doo-chul Choi, and Sang Woo Kim


Segmentation and Representation, Segmentation Evaluation, Pattern Recognition


A new unsupervised ensemble evaluation algorithm is proposed for image segmentation. A semi-supervised support vector machine is used to combine the existing unsupervised evaluators. We also proposed feature extraction and data selection procedures to enhance the overall performance. We experimentally demonstrated that our proposed algorithm is superior to existing segmentation evaluation measures.

Important Links:

Go Back