Webbapplikation för segmentering av hustak baserat på region growing

Sammanfattning: The need for energy is constantly increasing, and fossil fuels are both harmful and neither everlasting. This puts higher demand on the production of renewable energy. In Sweden the production of renewable energy is relatively good, however the solar power is not utilized enough. The planning phase for the purpose of solar installation is a lengthy process, and will over the coming years need to be more efficient. Several different methods are used for planning today, with the help of geographic data in the form of ortophotos. Multiple studies today have successfully achieved the segmentation of roofs, which then is used in this process. The studies that have achieved this, has used software that is not widely available to private individuals and thus the information regarding precise solar potential assesments are unavailable to them. This study has not found a similar implementation which operates in a semi-automatic manner as a webservice, on the Internet which is the most used platform. Therefore this study will investigate how roofs can be segmented and drawn on a map in a web application, in an automated manner with image processing. This is accomplished using the region growing algorithm. The result is produced by a number of tests on a variation of roof designs and colors. Validation of the results are done visually, when the segmentation is drawn onto the map. In most cases the algorithm has not been able to correctly identify and segment parts of the roof, and in no case has been successful to segment an entire roof. This was largely because of lack of data, and shadow disturbance in the images used. The result shows the method could work in a better way, using more data and developing the image processing further. It also shows how easily this type of method can be integrated with a webmap.