Combined Center-Symmetric Local Patterns for Image Recognition

Local feature description is gaining a lot of attention in the fields of texture classification, image recognition, and face recognition. In this paper, we propose Center-Symmetric Local Derivative Mapped Patterns (CS-LDMP) and eXtended Center-Symmetric Local Mapped Patterns (XCS-LMP) for local description of images. Strengths from Center-Symmetric Local Derivative Pattern (CS-LDP) which is gaining more texture information and Center-Symmetric Local Mapped Pattern (CS-LMP) which is capturing nuances between images were combined to make the CS-LDMP, and similarly, we combined CS-LMP and eXtended Center-Symmetric Local Binary Pattern (XCS-LBP), which is tolerant to illumination changes and noise were combined to form XCS-LMP. The experiments were conducted on the CIFAR10 dataset and hence proved that CS-LDMP and XCS-LMP perform better than its direct competitors.