Remote sensing

Objects in images

One of the most common processing requirements is the identification of objects and extraction of information regarding the feature of interest. Object based image analysis (OBIA) comprises image segmentation and classification steps in an attempt to produce meaningful maps of features in an image. These could be fields, buildings, roads, lakes etc. The accuracy of the object extraction is a function of image quality, landscape heterogeneity, and the segmentation algorithm being used. The algorithms generally require inputs related to scale, shape and compactness of the features within the image, and the choice of values for those parameters is made in an iterative manner. The following three images show the output from the original image, through a clustering stage to a segmented version of the original. The images contain a block of 3 houses, a road, some cars and vegetation in the rear gardens. The images shown are a subset of a larger image collected from a hot-air balloon in 2009.

a) The subset original image
a) The subset original image

b) Filtered and clustered
b) Filtered and clustered

c) Objects overlain on original image
c) Objects overlain on original image

Using actively developed open source software and custom workflows, Geoger Ltd. can effectively produce object based maps from a variety of image data sources. If you need to find and extract information about specific fields, houses, cars etc from imagery then get in touch. The objects can then be converted into GIS-ready vector formats to allow further analysis of the features of interest.

Better understand your data by contacting Geoger Ltd today!