One of the current research problems in the field of visual surveillance is optimizing the locations of the visual sensors. The efficiency of visual sensor localization becomes increasingly demanding as it can directly impact the efficiency of allocated resources as well as system performance. This problem can be divided to two sub-problems namely how to prioritize the area of interest and how to optimize the positioning of the cameras. This paper introduces a novel method for hazardous based area classification for visual surveillance purposes. The mapping process relies on per pixel weighting mechanism to evaluate the contribution of each pixel in the map plane in the overall significance of the area. This process is followed by proposing a combination of prior knowledge of the area with analyzing the generated edges to localizing the possible position set. The proposed technique is capable of locating the possible set of the visual sensor positions which is considered a primary requirement for the optimization of the camera placement problem. © 2014 IEEE.