Introduction to ERDAS IMAGINE

About This Training

ERDAS IMAGINE is a powerful remote sensing software suite used for viewing, processing, and analyzing geospatial data. In this tutorial, we will focus on four key functions: supervised classification, unsupervised classification, recode, and accuracy assessment.

Supervised Classification

Supervised classification is a method where the user provides the software with training samples of known classes to classify pixels in an image. Here's how to perform supervised classification in ERDAS IMAGINE:

• Open Image: Begin by opening the image you want to classify.
• Training Samples: Define training samples for each class you want to classify. These samples should represent the spectral characteristics of the classes accurately.
• Classifier Setup: Set up the classifier. Common classifiers include Maximum Likelihood, Minimum Distance, and Mahalanobis Distance.
• Classification: Run the classification process using the defined training samples and classifier.
• Post-classification Processing: Refine the classified image using tools like filtering and editing.

Unsupervised Classification

Unsupervised classification is a method where the software identifies clusters of pixels with similar spectral properties and assigns them to different classes. Here's how to perform unsupervised classification in ERDAS IMAGINE:

• Open Image: Open the image you want to classify.
• Clustering Algorithm: Choose a clustering algorithm such as K-means or ISODATA.
• Number of Classes: Specify the desired number of classes or let the software determine it automatically.
• Classification: Run the unsupervised classification process.
• Class Refinement: Refine the resulting classes if necessary by merging or splitting them.

Recode

Recode is a process where you assign new values to pixels based on their original values. It is often used to simplify data or create new thematic maps. Here's how to perform recode in ERDAS IMAGINE:

• Open Image: Open the image you want to recode.
• Recode Setup: Define the criteria for recoding, such as a range of pixel values or specific classes.
• Assign New Values: Specify the new values or classes to assign to the pixels.
• Run Recode: Execute the recode process.
• Review Output: Check the recoded image to ensure it meets your requirements.

Accuracy Assessment

Accuracy assessment is essential for evaluating the reliability of classified images. It compares the classified image against reference data to determine classification accuracy. Here's how to perform accuracy assessment in ERDAS IMAGINE:

• Prepare Reference Data: Gather reference data, typically through ground truthing or existing datasets with known classes.
• Classified Image: Have the classified image ready for comparison.
• Accuracy Assessment Setup: Set up the accuracy assessment tool in ERDAS IMAGINE.
• Comparison: Compare the classified image against the reference data to determine the number of correctly classified pixels.
• Calculate Accuracy Metrics: ERDAS IMAGINE provides various accuracy metrics such as overall accuracy, user's accuracy, and producer's accuracy.
• Interpret Results: Analyze the accuracy metrics to understand the performance of the classification.