Supervised and Unsupervised Classification

 Introduction to Unsupervised and Supervised Classification in ERDAS

In this week's lab we learned how to review and edit unsupervised and supervised classifications of aerial imagery. Through ERDAS Imagine you can set limits to the number of classes that each pixel could represent the categories. You can also reclassify the categories into to better summarize the image. 

Additionally, you can combine classifications by recoding the attributes and by doing so assigns a value that new categories are classified as. In our lab, after classifying each attribute to 5 classes we calculated the areas of each category. Since our data was using land classifications of forest, water, and agriculture areas we calculated the total area of permeable ( where water can be absorb into the soil such as vegetation, gravel, or turf) and impermeable ( where water is unable to be easily absorbs into the soil such as buildings, sidewalks, parking lots etc.). 

Supervised Classifications has a attribute table that already organized thematic raster data which takes less occupation space. Part of processing the data is to input the class type specifications into a process call Spectral Signatures that allows the Classifier to recognize the different features each data class has. We learned two forms of collecting spectral signatures through the Supervised Tab or through "Seed" a program that grows a region around specific geographic points of known land covers in the data set. 



Comments

Post a Comment

Popular posts from this blog

Fundamentals in Python

Image
  Flowchart Working with Spyder was a challenge. I had started on step 1 on a different approach than the lab instructions gave. My First Attempt started as follows stringName = “Brittany Anne Joe”  nameList – (“Brittany” “Anne” “Joe”)  print (“Joe”)  Although this gave the same results of printing my last name Joe in the program I learned to use some of Python’s functions that allows me to print my last name from a list. stringName.split automatically takes your stringName feature and divide it into separate cars for “Brittany” “Anne” “Joe”. To specify what to print from a list you need to index by specifying the number that represents your list. So “Brittany” would be 0, “Anne “would be 1, and “Joe” was 2.  After step 1, we had to debug a prewritten code that generates the results of random wins and losses for the players. I was challenged to create a loop to generate 20 random numbers between 0-10. My unlucky value was number 5 and we had to use an if/else statement to count the num
Read more

Module 5: Choropleth and Proportional Mapping

Image
 This week's assignment challenges us to explore data sets and using graduated and proportional symbology to present data. Graduated symbology utilizes classified data that shows the quantitative differences between data in which data is classified in ranges and a different sized symbol represents that range. Proportional symbology uses unclassified data to represent values in a field using the attribute value.  One challenge in this week's lab was working on the symbology by inputting a picture symbol into the mapping project. Another challenge to the week's lab is adjusting the European country's labels. Due to the close proximity of some countries or the size of the country the label wouldn't fit inside. Many adjustments were made, but there can be further improvements in future projects. 
Read more

Module 3: Cartographic Design

Image
 Gestalt's Principles   In our week's lesson, we learned about Gestalt's Principles covers cartographic design techniques that allows the audience to better understand a map. Few points about cartographic design and how gestalt's principles apply to map design is understanding the four techniques to display data.  Visual Heirachy Using symbols and ranking methodology to highlight data importance or where you want your audience to focus on in map. You can make single symbols different sizes to represent different classes or highlight their value on the map compared to other data such as population maps.  Contrast  Contrast focuses on differences between map features such as adjustments to size, color, or shape. Sometimes certain color choices such as blues and reds are great for highway and interstates. It can be helpful with a neutral background.  Balance  Balance is the organization of all map elements that fills the space granted for the map project, but use empty spa
Read more