The grouping of facial images using agglomerative hierarchical clustering to improve the CBIR based face recognition system
Muhammad Fachrurrozi, Clara Fin Badillah, Saparudin Saparudin, Junia Erlina, Erwin Erwin, Mardiana Mardiana, Auzan Lazuardi · 2017
The grouping of face images can be done automatically using the Agglomerative Hierarchical Clustering (AHC) algorithm. The pre-processing performed is feature extraction in getting the face image vector feature. The AHC algorithm performs grouping using linkage average, single, and complete method. Grouping face images can help improve the search speed of the CBIR based face recognition system. The cluster validation test uses the value of Cophenetic Correlation Coefficien (CCC). From the test results, it is known that the complete method has a higher CCC value than other methods, that is equal to 0.904938 with the difference value of 0.127558 on single method and the difference of 0.02291 on the average method. The face recognition system using pre-processing clustering can perform faster face recognition better than without pre-processing clustering.