This paper presents an image processing approach to detect obstacles on road using a monocular IR camera. Since potholes and speed bumps are the two major obstacles on roads, the proposed method focuses mainly on detecting them. Shadows are the major challenge in image processing because their presence in the image frame may lead to false results. A simple method to compensate shadows has also been proposed. Open Source Computer Vision (OpenCV), developed by Intel is used for implementation of the algorithm. The major target application area could be the powered two wheeler segment in emerging markets, where potholes and speed bumps might be a potential cause for accidents. Another application can be to upload the pothole information on to a cloud to provide intelligent navigation information via Global Positioning System (GPS).