The 3D building modeling is one of crucial components in constructing 3D geospatial information. However, most of the existing methods for 3D building modeling depend on manual procedures using photogrammetric and aerial laser scanning data. Such methods take huge amount of time and efforts and have a disadvantage of poor building façades description. On the other hand, many researchers have been interested in constructing 3D building models from point clouds acquired by terrestrial laser scanner. Similar to the aerial laser scanning data, the work process for 3D building modeling is also mainly dependent on the manual process which is inefficient in terms of time and cost. In this regards, this research proposes a semi-automated framework for 3D building modeling using terrestrial laser scanning data. Two key steps: 1) Segmentation based on RANSAC algorithm; 2) Outline extraction after rasterization, are proposed in this research. More specifically, optimization of four different parameters in the segmentation procedure is implemented. The regularization technique is applied to the outlines which are derived from the boundary tracing process. The performance of the proposed framework is evaluated by comparing the processing time between the proposed and fully manual one. Around 30% of the efficiency is improved when the proposed framework is applied for 3D building modeling compared to the manual process.