Novel human age estimation system based on DCT features and locality-ordinal information

The human face aging is a typical process, which the facial images have different face patterns in aging feature space. This paper present a Novel human age estimation technique using Discrete cosine transform and PLO for human age estimation. The technique shows the greatest experimental outcomes of with proposed method, as well as the corresponding optimum numbers of the selected features. Evaluate those methods on the basis of Mean Absolute Error (MAE). The MAE is utilized in most of the papers. MAE is described as the average absolute errors among predicted and real or actual age. The algorithm becomes more accurate when the MAE value is lowest. To improve the performance, we will consider the three data sets. We will applied the proposed technique on three different data sets namely Faces dataset, the FG-Net dataset and Images of Group dataset and results evaluted. Lastly after evaluting the performance of our algorithm we trained our classifier with data set and tested over random internet images the system is quite sucessful on internet images as well.