Insight Tools

Insight Tools provide a set of visual and algorithmic operations that can be applied directly to the insight currently displayed on the map. These tools allow you to refine visualization, enhance contrast, and extract patterns based solely on the active insight.

Overview #

Insight Tools operate exclusively on the data being visualized.

• No additional bands are used beyond the current insight.
• The behavior depends on whether the insight is a single-band index or a multi-band composition (RGB).

These tools are divided into two main categories:

• Visual Adjustments
• Algorithmic Analysis

Visual Adjustments #

Visual tools modify how the insight is displayed without changing the underlying data structure.

1. Stretching (Color Scaling) #

Stretching controls how values are mapped to colors, improving contrast and interpretability. It defines the visible range of the insight and helps emphasize meaningful variation in the data.

Available methods:

• Sigma-based:
  • 1 Sigma
  • 2 Sigma
  • 3 Sigma
  Sigma-based stretching uses the statistical distribution of pixel values around the mean. Lower sigma values create a narrower display range, increasing contrast and making subtle differences easier to detect. Higher sigma values include a broader range of values, preserving more of the overall variation in the insight.
• Percentile-based:
  • 90%
  • 98%
  • 100%
  Percentile-based stretching defines the visible range by excluding extreme low and high values based on their position in the distribution. This is useful for reducing the visual impact of outliers and improving the visibility of the most representative values in the image.

These methods adjust the visible range of values, helping to highlight relevant features in the data.

2. RGB Adjustments (Compositions Only) #

For multi-band (RGB) insights, additional visualization techniques are available.

Decorrelation Stretch #

Enhances color separation by reducing band correlation, making subtle differences more visible.

Enhanced RGB Contrast #

Two approaches are available:

• Equalization:
  • Define lower and upper percentiles.
  • Improves contrast distribution across the image.
• Gain-based adjustment:
  • Apply gamma correction to each channel.
  • Enhances contrast and visual differentiation.

These tools are especially useful for highlighting geological or environmental patterns in RGB compositions.

3. Color Palette (Single-band Indices) #

For single-band insights, you can modify the color representation using predefined palettes.

Available palettes include:

• Default
• Grayscale
• Viridis
• Inferno
• Magma
• Plasma
• etc

Changing the palette helps adapt the visualization to different analytical needs.

4. RGB Adjustments (Compositions Only) #

For single-band insights, you can also define a minimum threshold, a maximum threshold, or both.

Thresholding allows you to limit the visible range of the insight by removing values outside the selected interval:

• Values below the minimum threshold are excluded.
• Values above the maximum threshold are excluded.

The available threshold range is always constrained by the minimum and maximum values of the insight currently being visualized, and those limits are shown on top of the threshold selection area.

This operation does not create a new insight. Instead, it replaces the current visualization with the thresholded result so you can immediately evaluate how the mask would look on the map.

Once thresholds are set, a Save Mask option becomes available. This allows you to save the thresholded result as a custom mask.

These saved masks are considered private masks, meaning they are user-defined masks created from your own threshold settings. Once saved, they can be reused in future workflows with the same insight.

Algorithmic Analysis #

Algorithmic tools apply computational methods to extract structure and patterns from the insight and create new insights derived from the base one.

1. Edge Detection #

Edge detection highlights boundaries and transitions within the data.

Available algorithms:

• Zero-crossing: Detects edges based on second derivative changes.
• Canny: Produces clean and accurate edges using gradient intensity.
• Laplacian: Applies a second-order derivative to detect edges.
• Sobel: Computes gradient magnitude to emphasize edges.

These methods are useful for identifying structural features, boundaries, or anomalies.

2. Unsupervised Classification (Clustering) #

Clustering groups pixels based on similarity, without predefined labels.

K-means #

• You define the number of clusters (classes).
• The algorithm groups pixels accordingly.

X-means #

• You define the maximum number of clusters.
• The system automatically determines the optimal number of classes.

This is an unsupervised machine learning approach, commonly used for segmentation and pattern discovery.

Summary #

• Insight Tools operate only on the currently visualized insight.
• Visual adjustments improve interpretation without altering data.
• Algorithmic tools extract structure and patterns from the insight and create new insights.
• Available functionality depends on whether the insight is single-band or multi-band.

Each analytical method is documented in detail in its corresponding section.