Interest Point Explorer

The Interest Point Explorer allows you to visualize, analyze, and manage interest points (features) detected in your views. Interest points are used for image registration and stitching.

Overview

The explorer displays a table showing all interest point detections for the selected view(s). Each row represents a different interest point detection/label, showing statistics about the number of detections, correspondences, and which views contain them.

Table Columns

Interest Point Label: The name/identifier of the detection (e.g., "beads", "DoG")
#Detections: Total number of interest points detected
#Corresponding: Number of points that have correspondences to other views
#Correspondences: Total number of point-to-point correspondences established
Present in Views: List of ViewSetup IDs and Timepoints where this detection exists
Parameters: Detection parameters used (if saved)

Visualization Features

Point Size Control

Use the Point Size slider (top right) to adjust the size of displayed interest points in BigDataViewer:

Left (smaller values): Tiny points for detailed inspection
Center (value ~30): Default size
Right (larger values): Large points for easier visibility
Can also edit the size value directly in the text field next to the slider
Uses exponential scaling for fine control at small sizes

Plane Thickness Control

Use the Plane Thickness slider (top right) to control which points are visible based on distance from current plane:

Left (thin): Only show points very close to current viewing plane
Right (thick): Show points further away from current plane
Can also edit the thickness value directly in the text field
Useful for reducing visual clutter in 3D datasets
Uses power scaling (value^5) for intuitive control

Color Coding

Each interest point detection is assigned a unique color for visualization in BigDataViewer. Click on different rows in the table to highlight different interest point sets.

Interactive Table Features

Clicking on Cells

Left-click on table cells to visualize interest points in BigDataViewer:

First click on a cell: Shows all corresponding points (highlighted in red)
Second click on same cell: Shows only inter-view visible correspondences (highlighted in green)
Third click on same cell: Deselects (returns to default color)
Clicking different rows switches between different interest point detections

Context Menu Operations

Right-click on table rows to access additional operations:

Delete: Remove the selected interest point detection(s) from the dataset
Note: Deletion is immediate and cannot be undone without reloading the XML

Keyboard Shortcuts

F1	Show this help window.

Tips and Best Practices

Multiple Views Selected: When multiple views are selected in the ViewSetup Explorer, the Interest Point Explorer shows detections common to all selected views
Correspondence Analysis: Compare #Corresponding vs #Detections to assess registration quality - low ratios may indicate poor feature matching
Performance Mode: For large datasets with many points, reduce the Plane Thickness to only display points near the current viewing plane
Color Management: Each detection gets a unique color - use the table to quickly identify which points belong to which detection in BigDataViewer
Rename Before Processing: Give detections meaningful names before running registration/stitching to make it easier to track which parameters were used

Common Workflows

1. Inspecting Detection Quality

Select a single view in the ViewSetup Explorer
Open the Interest Point Explorer
Click on a detection row to visualize points in BDV
Adjust Point Size and Plane Thickness for optimal viewing
Check if points are well-distributed and located on actual features

2. Analyzing Correspondences

Select two or more views in the ViewSetup Explorer
Click on a detection row
Click on the #Corresponding or #Correspondences column to highlight corresponding points
Click again to show only inter-view visible correspondences (green)
Verify that correspondences are correct by visual inspection

Technical Details

Point Visualization

Distance Fade: Points fade based on distance from current plane (fade factor: 0.125)
Size Scaling: Exponential scaling formula: size = base_size × 10^((slider-30)/85)
Plane Thickness: Power scaling formula: thickness = 100 × (slider/100)^5

Data Structure

Interest points are stored per view (ViewId = TimePoint + ViewSetup)
Each detection has a label/identifier
Correspondences link points across different views
Both detections and correspondences are saved in the XML metadata