Human Age Estimation Using Biologically Inspired Features

In the public security, criminal investigation, electronic customer rela-tionship management, network access control, e-commerce and other ?elds,the age estimation have important practical applications.There have beenmany proposed means of age estimation.In this paper, we adopted thebionic features to study age estimation based on face images. The mainwork is as follows:1. This article has been improved the bionic features.Usuallly, we or-ganize the S1 layers of the image into di?erent band according to the sizeof Gabor ?lters in bionic feature model.We divided the images of S1 layerinto several overlapping blocks.In each band, we operate MAX on orga-nized a?erent S1 units from the previous layer with the same orientation.In every block we take MAX operator again so we can get C1 layer.In ourimproved model,we selected block by the key point which is considered asthe center and then take STD operator in each block. This greatly reducesthe time of extracting C1 features of the whole image, and increase thee?ective utilization information of the key points.2. We proposed one new age estimation method that based on improvedbionic feature model and SVM.The basic process is: First, the necessarypretreatment of the image, then feature extraction, dimensionality reduc-tion and SVM classi?cation of face images to estimate the subject's age.3. We proposed this method of age estimation based on the improvedbionic features and PLS. This method need not be more down-dimensionalpart and it can be directly applied on the extracted features.4. We programmed the above algorithm and veri?ed the algorithm onthe FGNET database. Experimental results show that our algorithm hassome advantages compared with existing methods.