Abstract for presentation at The 13th Australasian Remote Sensing and Photogrammetry Conference

Fusion of multi-sensor data is becoming a widely used procedure since the availability of complementary nature of dissimilar datasets. The combined use of high spatial resolution imagery and lidar derived digital surface model (DSM) can reduce interclass confusion in the classification process. However, pixel-based data fusion in a multi-dimensional feature space does not make use of any spatial concepts; the DSM is just used as an additional image layer. Object-based fusion overcomes this shortcoming by segmenting multi-source images into meaningful multi-pixel objects of various sizes, based on both the spectral and spatial characteristics of groups of pixels. This paper compares the results of the pixel- and object-based fusion of a lidar derived DSM with colour aerial photography and multispectral imagery. The comparison is based on the assessment of the classification accuracy where reference information collects through field survey. Pixel-based classification of the colour photography and the DSM exhibits better results than solely using colour photography. The same result is found for the multispectral imagery and the DSM. Object-based fusion achieves superior results compared to all pixel-based classification of tested categories. Object-based fusion of the colour photography and the DSM shows the highest classification accuracy (83.6 percent). Multispectral imagery and the lidar derived DSM achieve 81 percent classification accuracy. These results imply that the high spatial resolution of colour photography have a greater influence on the fusion process than the spectral and radiometric resolution of the multispectral imagery.