ElectroEncephaloGram (EEG) signals based localization of brain sources has been an active area of research. The localization accuracy is limited by the assumption of head shape. Spherical and hemispherical head models have been studied in the literature. The corresponding basis functions that include spherical harmonics and hemispherical harmonics have been utilized to represent the EEG potential. In this work, a new set of basis functions (called head harmonics) dedicated to represent potential over head are introduced. Subsequently, MUltiple SIgnal Classification (MUSIC) and Minimum Variance Distortionless Response (MVDR) methods have been reformulated in head harmonics domain. The Three Layer Concentric Spherical Head (TLCSH) model is utilized for this purpose. The advantage of applying head harmonics in terms of computation efficiency and localization accuracy is demonstrated using various simulation experiments. A relative performance of the source localization algorithms is detailed. The proposed data model and localization algorithms has been additionally, verified for real EEG data.