Panca Mudjirahardjo, M. Fauzan Edy Purnomo, Rini Nur Hasanah, Hadi Suyono
Feature extraction plays an important role in head recognition. It transforms an original image into a specific vector to be fed into a classifier. An original image cannot be further processed directly. Raw information in an original image does not represent a specific pattern and a machine cannot understand that information. In this paper, we evaluate the multi scale performance of feature extraction method for human head recognition. We perform a comparison of the existing image features extraction and our novel methods using a static image. The existing features are HOG and LBP, and the novel feature is a histogram of transition. To extract the image feature, we use a normal size of image weight and height of 20 and 30 pixels, respectively. Then we evaluate the image with size of less than and greater than the normal size. We employ an algorithm to increase or reduce the image size into the normal size. The recognition rates using the proposed feature are that the head recognition rate is 91% and the non-head recognition rate is 99.7%. The execution time is 0.077ms. These performances show that the proposed feature can be used for real time application. © Research India Publications.
Department of Electrical Engineering, Universitas Brawijaya, Jl. MT. Haryono 167, Malang, 65145, Indonesia
