Showing posts with label [GIS 5027] Aerial Photography and Remote Sensing. Show all posts
Showing posts with label [GIS 5027] Aerial Photography and Remote Sensing. Show all posts

Module 5 - Unsupervised & Supervised Classification

        This week's module involved performing unsupervised and supervised classification using ERDAS Imagine. We were provided satellite imagery and created a land use classification using spectral signatures. Additionally, I generated signatures for roads and water, but due to spectral overlap between road and deciduous forest signatures, I had to create additional signatures to address this issue.


Module 4 - Spatial Enhancement, Multispectral Data, and Band Indices

     This week's module delved further into the features of ERDAS Imagine. We learned how high-pass and low-pass filters change aerial imagery, how image histograms are used to assist in feature identification, and how different band combinations enhance specific features such as water bodies.

    This map layout depicts an area of Washington State where a stream meets the bay. Highlighted in light blue is the feature of interest that is bright in band layers 1-3, and unchanged in layers 5-6. Using a combination of Red: 4, Green: 3 and Blue: 2 I was able to highlight this variation in the water.

Module 3 - Intro to Electromagnetic Radiation (EMR), Sattelite Sensors and Digital Image Processing.

     This week's module introduced us to ERDAS Imagine and it's features. Specifically, we were tasked with creating a subset of an aerial image in ERDAS to use as a base for a map layout in ArcGIS.


      This map represents the land classification of a subset of Washington State by hectares. This map frame was clipped from within ERDAS.

Module 2 - Land Use Land Cover Classification

     This week's module involved digitizing an area of Pascagoula, Mississippi. We had to classify aerial imagery of the area by land use or land cover based on features identified. The classification used was the USGS Standard Land Use / Land Cover classification system and the imagery was classified up to Level II. The creation of this map involved heavy use of Clip within the Modify Features pane of the Edit tab.

    The second part of the module involved testing the newly classified imagery using ground truthing. To achieve this, I chose 30 random sample points and checked their land use codes against information found using Google Street View. Most of the inaccuracies were due churches or small grocery stores being classified as 11 - Residential instead of 12 - Commercial and Services due to their location within larger residential areas.


Module 1 - Visual Interpretation

     The objective of this first module was to teach us how to interpret aerial imagery based on criteria such as Tone (brightness), Texture, and recognition elements such as Shape Size, Shadow, Pattern and Association. These criteria help to distinguish features on aerial imagery. The map layout below shows different tones and textures found within aerial imagery. The layout was made without a north arrow or scale bar due to the raster image having an unknown coordinate system.