Modified chess patterns: handcrafted feature descriptors for facial expression recognition

Abstract Facial expressions are predominantly important in the social interaction as they convey personal emotions of an individual. The main task Expression Recognition (FER) systems is to develop feature descriptors that could effectively classify facial into various categories. In this work, towards extracting distinctive features, Radial Cross Pattern (RCP), Chess Symmetric (CSP) and (RCSP) have been proposed implemented a 5 $$\times $$ × overlapping neighborhood overcome some limitations existing methods such (CP), Local Gradient Coding (LGC) its variants. neighborhood, 24 pixels surrounding center pixel arranged two groups, namely which extracts values by comparing 16 with one value from remaining 8 pixels. experiments conducted using RCP CSP independently also their fusion RCSP different weights, on variety expression datasets demonstrate efficiency methods. results obtained experimental analysis