There has been a growing research interest in 3D and free viewpoint video systems to enrich the immersive viewing experience of multimedia users. To efficiently support such video transmissions, a multi-view video plus depth coding architecture has recently been introduced using depth maps. Depth maps have distinct spatial and temporal characteristics and existing coding techniques are unable to exploit these characteristics fully in coding, which also affects the image quality of depth-image-based rendering. This thesis aims at designing an independent and autonomous depth map coder capable of exploiting unique depth characteristics to achieve high compression efficiency with better depth quality.