The aim of the work is to create methods and software tools for the terahertz range intelligent video surveillance, that is, for automatic analysis of video images in the terahertz frequency range. Terahertz video surveillance provides unique prospects in the area of public safety; in particular, it allows you to remotely identify weapons and dangerous items hidden under clothing on the human body. However, such characteristics of terahertz video, as low resolution, low contrast, and low signal-to-noise ratio, lead to the need to develop new methods and approaches to automatic video analysis. To understand the terahertz video image, the operator of the industrial video surveillance system usually compares it with images in other frequency ranges (visible or infrared). The comparison of these images of different types helps the operator to interpret the colored and/ or one-color spots in the terahertz image in a proper way. In terms of automatic video analysis, this means that the context of the observed events and objects is taken into account, or in other words, a semantic fusion is implemented of the terahertz range video image with video images of other frequency ranges, e.g., near-infrared, visible, etc. The authors consider the semantic fusion of the video images as a critical component of the prospective terahertz intelligent video surveillance technology. A new method of the terahertz video surveillance based on the fusion of the terahertz video with 3D video is proposed. The means of the object-oriented logic programming developed for the semantic fusion of the terahertz and 3D video images are described. The developed method provides a real-time fusion of the terahertz video acquired using the THERZ-7A (Astrohn Technology Ltd) subterahertz scanning device (0.23-0.27 THz) and 3D video data acquired using the Kinect 2 (Microsoft Inc) time-of-flight sensor. The method and software tools for the semantic fusion of the terahertz and 3D video images are developed.