pixelcooker

Image classification for landcover mapping has so many types of algorithms or methods. One of them is Image Segmentation and Classification. The process can be described as imagery is segmented into many segments, and from there we classify them according to the specified classification schema.  This video tutorial below is about how to perform object-based image segmentation and classification using Satellite Imagery data. Not only the segmentation, but I also made this video as a workflow, which means I demonstrate from the training samples generation (for supervised classification), ground truth creation data to support accuracy assessment, the image segmentation process, and closing with the accuracy assessment to inspect the accuracy of the final classified raster. Almost every step is automatic so it will save a ton of time and effort.  So if you are curious, check this below

All remote sensing users must be familiar and used to playing with multispectral bands combination to get certain information from imagery. If you are deeply understand about spectral response curve of surface earth objects, you will know that certain object (vegetation for example) gaves different spectral response compared to another objects. For satellite imagery consist of more than 3 bands (RGB in digital camera), different bands combination in RGB planes gaves different emphasizes to landcover colors in the imagery. Hence they are called multispectral imagery. Most of low to medium resolution imaging satellite currently operational in the earth orbit are multispectral, like landsat, modis, noaa avhrr, and many others.  Multispectral imagery number of bands can range from 4 bands to dozens bands, but not reaching hundreds. When the bands number reached more than one hundred bands it is called hyperspectral imagery and its analysis and image processing workflow would be pretty ...