Application of template matching method to traffic feature detection using KOMPSAT EOC imagery

Abstract

Currently, the application domains of the commercialized high resolution satellite imagery are in the middle of expansion. These images contain detailed information to investigate targets in the application level. In this paper, we try to demonstrate traffic analysis scheme based on template matching method algorithm. The simple matching algorithm implemented for this purpose is as follows: First, one user determine a type of vehicle or other traffic feature, as a template corresponding to a target feature. In this process, it is possible to choose multiple templates according to vehicle modes. Second, one can automatically detect and count the number of vehicles statistically similar to template given in the previous step through template matching method algorithm. While, if user already knows the geo-referenced location of a specific vehicle, traffic flow analysis in the static status is also possible in this step. Through the results from each counting vehicles, user can summarize statistics of traffic features at a certain specific time in high-resolution satellite imagery. As for the actual test and experiment, satellite imagery of KOMPSAT EOC was used, and KOMPSAT imagery consists of feature sets in an urban region showing complex types of spatial features. Results of template matching method can provide useful ones in detecting the types of vehicle in the remote traffic detection or monitoring system. In conclusion, this approach can be utilized to obtain some information about on-road traffics such as the location of vehicle, traffic volume, and other transportation parameters, related to transportation uses of remote sensing image.