Road obstacles cause serious accidents that have a severe impact on driver safety, traffic flow efficiency and damage of the vehicle. Detecting obstacles are important to prevent or to reduce such kind of the accidents and fatalities. However, it is difficult and becomes tricky because of some problems like presence of shadow, environmental changes or a sudden action of any moving things (e.g., car overtaking, animal coming) and many more. Thereby, this paper aims to design an obstacle detection technique based on (i) moving cameras and (ii) moving objects. These methods are applied to obstacle detection phase, in order to identify the different obstacles (e.g., potholes, animals, stop sign, obstacles, bumps, road cracks) by considering road dimensions. A new technique is introduced for detecting obstacles from moving camera and moving objects which overcomes several limitations over stationary cameras and moving/stationary objects. Further, paper reviews recent research trends to detect obstacles for moving cameras and moving objects with discussion of key points and limitations of each approach. Finally, the results show that the proposed method is more robust and reliable than the previous approaches based on the stationary cameras.