Robust Active Shape Model via Hierarchical Feature Extraction with SFS-Optimized Convolution Neural Network for Invariant Human Age Classification

摘要: The features and appearance of the human face are affected greatly by aging. A is an important aspect for age identification from childhood through adulthood. Although many traits used in estimation, this article discusses classification using salient texture facial landmark feature vectors. We propose a novel (HAC) model that can localize points face. robust multi-perspective view-based Active Shape Model (ASM) generated achieved Convolution Neural Network (CNN). HAC subdivided into following steps: (1) at first, detected aYCbCr color segmentation model; (2) localization done on connected components approach ridge contour method; (3) three-sided polygon meshes perpendicular bisection triangle; (4) extraction anthropometric model, carnio-facial development, interior angle formulation, wrinkle detection heat maps; (5) Sequential Forward Selection (SFS) to select most ideal set features; (6) finally, (CNN) classify according correct group. proposed system outperforms existing statistical state-of-the-art methods terms accuracy, achieving 91.58% with Images Groups dataset, 92.62% OUI Adience dataset 94.59% FG-NET dataset. applicable research areas including access control, surveillance monitoring, human–machine interaction self-identification.