CCTV surveillance systems are widely used as a street monitoring tool in public and private areas. This paper presents a novel approach of an intelligent surveillance system that consists of adaptive background modelling, optimal trade-off features tracking and detected moving objects classification. The proposed system is designed to work in real-time. Experimental results show that the proposed background modelling algorithms are able to reconstruct the background correctly and handle illumination and adverse weather that modifies the background. For the tracking algorithm, the effectiveness between colour, edge and texture features for target and candidate blobs were analysed. Finally, it is also demonstrated that the proposed object classification algorithm performs well with different classes of moving objects such as, cars, motorcycles and pedestrians.