Lab 5 [GIS5027L]: Unsupervised and Supervised Classification


This week's lab focused on using the tools within ERDAS Imagine to perform unsupervised and supervised classification of pixels representing various features in an image. In the map above, supervised (maximum likelihood) classification was used to classify pixels based on eight different classes of land use.

The results above demonstrate many urban/residential areas were misclassified within the roads class. The results can be improved by selecting more signature samples for these two classes and then also evaluating these signatures to determine the optimal 3 bands that help with differentiation prior to running the classification algorithm. Two helpful tools in ERDAS Imagine for signature evaluation include examining histogram plots and mean plots of the image bands of the features you are seeking to differentiate.

Comments

Post a Comment

Popular posts from this blog

Lab 2 [GIS5935]: Data Quality Standards

Image
  The image above shows 20 test point locations across a street map for the city of Albuquerque. In selecting these test points to report the horizontal accuracy of two alternative geographic data sources providing street centerlines, the following constraints were applied, in compliance with using the National Standard for Spatial Data Accuracy to measure and report the associated quality (Minnesota Planning Land Management Information Center, 2019): Locations were selected using street intersections derived from orthophotos of the study area - these images contained detailed imagery of street centerlines and nearby features (e.g., medians, sidewalks) that could be readily used to accurately determine intersection points. Given the street uniformity throughout the region, approximately 25% of the points were derived from each quadrant of the overall study area, and all quadrants were equally represented in this way (meeting the requirement to draw at least 20% of test points from each
Read more

Lab 5 [GIS5935]: Surface Interpolation

Image
  In this lab we examined various interpolation methods to create a surface of water quality in Tampa Bay. Each method received as input a set of sample points containing Biochemical Oxygen Demand (BOD) values in milligrams per liter that were recorded from various points throughout the study region. The interpolation methods used this case study included the following: Thiessen IDW Spline - Regularized Spline - Tension This interpolation method is a special case of IDW where only the nearest water sample is used to estimate the BOD value at an unsampled location. This interpolation method uses a preset number of nearest neighbors to estimate the BOD value at an unsampled location, and inversely weights the contributing strength of these neighbors based on their distances from the unsampled location. Spline interpolation methods work to fit a smooth surface exactly touching the BOD sampled points, while
Read more

Final Project [GIS6005]: Conservation in the State of Florida: Preserving Key Habitats of the Striped Burrfish

Image
Striped burrfish are inshore fish known to exploit a wide range of depths. Although commonly found in grass beds around barrier islands in the Gulf of Mexico (Franks et al., 1972), they are also found near coral reefs (Robins, Ray, & Douglas, 1986), as well as in deeper waters during the winter months (Audubon Society, 2002). Changes to the geographic ranges of fish species due to factors such as ocean warming (Espino et al., 2019) are of concern to scientists. In addition, impacts to fish habitats in tourism-based states such as Florida are considered by economists given the socioeconomic benefits of commercial and recreational fishing to coastal communities (Lellis-Dibble, McGlynn, & Bigford, 2008). As one example of their impact, half of all federally managed fisheries in the U.S. depend on coral reefs (Burton, 2019). In addition, t he total value of coral reefs for southeast Florida was estimated in 2007 to be $174 million/year (Brander & van Beukering, 2013),
Read more