Real-time traffic flow analysis using road mounted surveillance cameras presents multitude of benefits. In this paper, we used surveillance videos to design optical flow based technique for robust motion analysis and estimation. Region growing method is employed for detection of objects of interest. Autonomous density estimation of vehicles is crucial for traffic congestion analysis so that countermeasures can be taken at the earliest possible opportunity. A video based data extraction scheme for traffic data is proposed to determine the right traffic conditions which alleviates the false alarms and detrimental noise effects. Evaluation of proposed system is done by applying approach on several surveillance videos obtained from different sources and scenarios. An experimental study illustrates estimation and analysis results accuracy as compared to state-of-the-art approaches.