DEVELOPING A METHOD TO GENERATE INDOORGML DATA FROM THE OMNI-DIRECTIONAL IMAGE

Abstract. Recently, many applications for indoor space are developed. The most realistic way to service an indoor space application is on the omni-directional image so far. Due to limitations of positioning technology and indoor space modelling, however, indoor navigation service can’t be implemented properly. In 2014, IndoorGML is approved as an OGC’s standard. This is an indoor space data model which is for the indoor navigation service. Nevertheless, the IndoorGML is defined, there is no method to generate the IndoorGML data except manually. This paper is aimed to propose a method to generate the IndoorGML data semi-automatically from the omni-directional image. In this paper, image segmentation and classification method are adopted to generate the IndoorGML data. The edge detection method is used to extract the features from the image. After doing the edge detection method, image classification method with ROI is adopted to find the features that we want. The following step is to convert the extracted area to the point which is regarded as state and connect to shooting point’s state. This is the IndoorGML data at the shooting point. It can be expanded to the floor’s IndoorGML data by connecting the each shooting points after repeating the process. Also, IndoorGML data of building can be generated by connecting the floor’s IndoorGML data. The proposed method is adopted at the testbed, and the IndoorGML data is generated. By using the generated IndoorGML data, it can be applied to the various applications for indoor space information service.