SmartTrace: Intelligent Boundary Detection Tool

A Java-based interactive tool for selecting and extracting regions from images with multiple selection methods.

Technical Overview

This application implements different selection algorithms to create precise selections in images:

• Point-to-Point Selection: Simple straight-line connection between selected points
• Intelligent Scissors: Edge-aware selection that finds optimal paths along image boundaries
  • CrossGradMono: Grayscale gradient analysis for edge detection
  • ColorWeight: Enhanced color-based edge detection for similarly bright but differently colored regions

Core Components

Selection Models

• Abstract SelectionModel defining selection state management and interactions
• PointToPointSelectionModel for straight-line selections
• ScissorsSelectionModel for edge-following selections

Graph-Based Path Finding

• Image pixels represented as graph vertices with weighted edges
• Dijkstra's algorithm implementation for optimal path detection
• Custom HeapMinQueue priority queue implementation using binary heap + hash table
• Asynchronous path computation with progress reporting

UI Components

• ImagePanel: Main component for displaying images and handling selection overlay
• SelectionComponent: Interactive overlay for selection interactions
• SelectorApp: Main application class managing UI components and user interactions

Key Implementation Details

|-- graph/
|   |-- Edge.java                    # Interface for directed edges in a graph
|   |-- Graph.java                   # Interface for directed graph structure
|   |-- HeapMinQueue.java            # Efficient priority queue implementation
|   |-- MinQueue.java                # Priority queue interface
|   |-- PathfindingSnapshot.java     # Captures state of pathfinding progress
|   |-- RefMinQueue.java             # Reference priority queue implementation
|   |-- ShortestPaths.java           # Dijkstra's algorithm implementation
|   |-- Vertex.java                  # Interface for graph vertices
|   |-- Weigher.java                 # Interface for edge weight functions
|-- scissors/
|   |-- ImageGraph.java              # Graph representation of image pixels
|   |-- ImagePathsSnapshot.java      # Visualization of pathfinding progress
|   |-- PolyLineBuffer.java          # Utility for building polylines
|   |-- ScissorsSelectionModel.java  # Intelligent selection implementation
|   |-- ScissorsWeights.java         # Edge weight functions for image features
|-- selector/
|   |-- ImagePanel.java              # Component for displaying the image
|   |-- PointToPointSelectionModel.java # Simple line segment selection
|   |-- PolyLine.java                # Immutable path of line segments
|   |-- SelectionComponent.java      # Overlay for selection interaction
|   |-- SelectionModel.java          # Abstract selection model
|   |-- SelectorApp.java             # Main application class

Technical Features

• Concurrent processing using SwingWorker for non-blocking UI
• Property change propagation for model-view communication
• Binary heap + hash table data structure for O(log n) priority queue operations
• Visual feedback during pathfinding with frontier/settled pixel visualization
• Interactive control point manipulation for selection refinement
• Image processing using Java's BufferedImage and Raster APIs

Requirements

• Java 17 or higher
• Swing-compatible environment

Implementation Notes

• The intelligent scissors implementation is based on graph algorithms with image gradient analysis
• Edge weights are calculated using perpendicular gradient analysis at pixel boundaries
• Binary heap with hash table indexing enables efficient priority updates for Dijkstra's algorithm
• Background thread processing with progress reporting for responsive UI
• Custom polygon extraction with alpha channel for saving selections