Terahertz (THz) communication enables ultra-highspeed wireless data transfer, especially for near-field communication (NFC). This study focuses on joint angle-of-arrival (AoA) and range estimation, crucial for near-field scenarios, using short dipole antennas model instead of point source one. Point source antennas, which can not be fabricated, are assumed in existing works for NFC, while short dipoles are practically fabricated and they behave as point sources in the far field. Therefore, assumption of point source antenna structure deviates significantly from the actual behavior of the antennas in the near field, which limits the existing solutions for NFC so far. Exploiting the complex electromagnetic theory of short dipole antenna, we propose a framework for AoA and range estimation that incorporates this short dipole antenna characteristic. Maximum likelihood (ML) based estimation is proposed, which combines an iterative method for high precision and computational efficiency along with the search space segmentation approach recursively. Numerical simulations demonstrate improved mean square error (MSE) and symbol error rate (SER) performance compared to the existing solutions based on the inaccurate point source assumption, proving its viability for near-field THz systems.