Sea-ice observation and estimation of ice forces are becoming increasingly important with the increased activities in arctic water. Proper modelling of ice properties and forces is crucial in such cases to ensure safe operations, for example, Dynamic Positioning (DP). This work, therefore, aims at developing algorithms for image processing to extract useful ice properties and subsequently aid the modelling of ice load exerted on a floating platform or a ship. A robust algorithm capable of detecting closely connected and overlapped ice floes with various sizes and shapes is presented, which is the first step for accurate modelling of ice forces. To demonstrate the effectiveness of this approach, image frames processed from videos produced during an experiment using a model ship performed at the National Research Council’s Ocean Coastal and River Engineering Research Centre (NRC-OCRE) in Canada are used. Simulated ice floe images are also used to show the efficiency of the proposed model and compare it with other published work. The model will be extended further to extract and correlate other ice properties with the ice forces and develop a machine learning based ice force prediction model. The predicted force from that model will then be used as a feedforward to Dynamic Positioning (DP) controller with an aim to improve the performance of the controller.