Skip to main content

Regression Models

Overview​

This model predicts a continuous value for each pixel in an image. This is ideal for cases where a user wants to predict a continuous value for each pixel in an image, such as a reflectance value, or a chlorophyll content.

Use Case​

Predict a continuous value for each pixel in an image.

How It Works​

Regression models are focused on predicting continuous outcomes from the spectral data, such as estimating chemical concentrations, vegetation indices, or temperature. These models analyze the spectral signatures across the hyperspectral image to establish a relationship between the spectral data and the continuous variable of interest.

Strengths​

  • Powerful in quantitatively predicting spatially varying parameters that are directly related to the spectral information
  • Enable precise estimation of physical, chemical, or biological properties across a hyperspectral image
  • Provide detailed and quantitative analysis of the scene

When to Use​

Best applied in scenarios where the objective is to measure or predict a continuous variable across an area, such as:

  • Assessing crop health through chlorophyll content
  • Predicting pollutant levels in water bodies
  • Estimating soil moisture content
  • Measuring vegetation stress indices

These models are ideal when you need to quantify specific attributes or conditions that can be directly correlated with spectral data, providing a detailed and quantitative analysis of the scene.