Surveillance systems have become increasingly popular in the globalization process. Intelligent video surveillance system based on image recognition is widely used to effectively prevent many crimes and helps to provide public security. Due to the high complexity in techniques such as real time processing and image contents analysis/understanding, a well-developed product is not available until now in this field. In this work, human presence is detected from the surveillance video by extracting the skin region (if present) from the selected frame. Since human skin can be found in a varied color, fuzzy approach is used to extract the skin region. The experimental result shows the effectiveness of the proposed technique.