*Author for correspondence Indian Journal of Science and Technology, Vol 9(31), DOI: 10.17485/ijst/2016/v9i31/93481, August 2016 ISSN (Print) : 0974-6846 ISSN (Online) : 0974-5645 Obstacle Detection from Still Images using Improved Background Subtraction Method Nishchala Thakur* and Vikas Wasson CSE Department, Chandigarh University, Mohali - 140413, Punjab, India; nishchalathakur@gmail.com, vikaswasson.cse@gmail.com Keywords: Background Image, Background Subtraction, Gamma Correction, Illumination, Obstacle Detection Abstract Background/Objectives: We propose an approach that aims at improving the current method of obstacle detection so as to improve efficiency, detection rate and running time. Methods/Statistical Analysis: Background subtraction is a technique used in obstacle detection to find out the area of hindrances for traversal or navigation. In this technique two consecutive frame of video is taken for differencing and resultant giving the area in which obstacle is identified. Findings: The current technique has some limitations working under high intensity and change of illumination. We propose a hybrid approach which include merging of gamma correction with background subtraction. This method reduces the level of illumination and works well at different weather conditions. Our proposed method is useful in detection of area of changing objects and would be useful in video surveillance applications. 1. Introduction With the swiſt advancement in computer science, detec- tion technologies in computer vision are enhanced more extensively in different aspects. Object detection is widely used in computer vision for public security, visual sur- veillance, robot path traversal, blind navigation system, object recognition, pedestrian detection systems and many other fields. Obstacle detection is one of the fun- damental areas of problem in mobile robot navigation. For navigation in the world, it is crucial to identify that portion of the world which are difficult or impossible for traversing. Popular sensors used in range-based obsta- cle detection systems include ultrasonic sensors, stereo vision, laser rangefinders, and radar as these sensors mea- sure the distance of obstacles to the robot, they are suited for the tasks of obstacle detection and obstacle avoidance. e major disadvantages of using such sensors is there interference with the environment which lead to difficulty in interpretation of output signals, high consumption of power consumption, high price of acquisition, poor reso- lution and inability to detect small obstacles. is signify the need of using visual method for obstacle detection. However, object detection technology still has many shortcomings such as slow detection rate, low detec- tion accuracy and poor environmental adaptability, false detection etc. us, object detection technology tends to have need in-depth study. Obstacle detection is an attempt to reveal the objects that are present in the locomotive path to prompt col- lision free navigation. e primary purpose of this technique is to utilize visual perception of machine to make it understand the structure of environment around it and to predict the hindrance in path by itself without any human intervention. Till now, various approaches are being proposed but fail to give ideal results. So, it will not be immoral to state that there is scope of improvement in them. In this paper, we have presented a review on the various obstacle detection techniques that can be used in different area and application. Background subtraction method, also known as frame differencing method is used for detection of obstacle from the video sequence. In this approach two consecutive frames with same background from video is taken and