|
This paper discusses the methodology of the implementation of an inventory of vegetation in an urban area using photogrammetric data in the form of color NIR "true - orthophotomap" (true - ortho) and the digital surface model (DSM) created with data from airborne laser scanning, or alternatively, with an automatic correlation of images. The vegetation inventory was conducted by classification on the basis of the characteristics contained in pixels of georeferenced true - ortho while taking into account the elevation data in the form of gridded DSM. To carry out the classification Erdas Imagine software was used. The correct classification process was preceded by the creation of the input data for this task. This data was obtained from the processing of digital aerial photos taken by a Vexcel UltraCam camera with the ground resolution GSD = 10cm and point clouds acquired from ALS. This processing included the generation of digital terrain model in the SCOP++ environment and the digital surface model in an Opals and Inpho environment. The Comparison of DSM created from two different sources of data showed the overall consistency and uniformity and the ability to use both models to generate a true - ortho product from digital aerial photographs. The work was performed on an INPHO photogrammetric workstation. "True - ortho" was generated from both the black and white NIR images and colour images. The classification carried out with the Erdas Imagine software proved that this software is suitable for classification based on the features extracted from the pixels with the simultaneous analysis of elevation data. Simultaneous use of data both from airborne laser scanning and colour infrared images made it possible to make an exact classification of vegetation on very difficult terrain, like built up urban areas. The results of the classification accuracy were evaluated by the visual verification in Google Street View application. At a time when airborne platforms are equipped by both sensors, i.e. high resolution digital camera and laser scanner, data fusion is a commonly used approach. This makes it possible to combine the advantages of both types of data, and carrying out an inventory of the vegetation in the town area is one of many possible applications of the combined data from ALS and CIR.
|