io.github.jspinak.brobot.analysis.compare (Brobot 1.1.0 API)

package io.github.jspinak.brobot.analysis.compare

Image comparison and analysis algorithms.

This package provides utilities for comparing images, extracting contours, and analyzing visual differences. These tools support advanced matching operations by enabling detailed comparison of visual elements beyond simple template matching.

Core Components

ImageComparer - Comprehensive image comparison with multiple metrics
ContourExtractor - Extracts and analyzes object contours from images
SizeComparator - Compares dimensions and sizes of visual elements

Comparison Techniques

Pixel-Level Comparison


 // Compare two images pixel by pixel
 ImageComparer comparer = new ImageComparer();
 ComparisonResult result = comparer.compare(image1, image2);

 double similarity = result.getSimilarityScore();
 Mat diffMap = result.getDifferenceMap();
 List<Rect> changedRegions = result.getChangedRegions();

Contour-Based Analysis


 // Extract contours from classification results
 ContourExtractor extractor = new ContourExtractor();
 List<MatOfPoint> contours = extractor.extract(binaryImage);

 // Filter by size and shape
 List<MatOfPoint> filtered = extractor.filterContours(
     contours,
     minArea,
     maxArea
 );

 // Convert to bounding boxes
 List<Rect> boundingBoxes = extractor.toBoundingBoxes(filtered);

Size-Based Comparison


 // Compare element sizes
 SizeComparator sizeComp = new SizeComparator();

 boolean sameSize = sizeComp.areSimilar(rect1, rect2, tolerance);
 double sizeRatio = sizeComp.getSizeRatio(rect1, rect2);
 boolean aspectMatch = sizeComp.aspectRatiosMatch(rect1, rect2);

Comparison Metrics

Structural Similarity (SSIM) - Perceptual similarity metric
Mean Squared Error (MSE) - Pixel-wise difference measurement
Histogram Comparison - Statistical color distribution analysis
Edge Comparison - Structural element matching
Contour Matching - Shape-based similarity

Use Cases

Change Detection


 // Detect UI changes between states
 ImageComparer comparer = new ImageComparer();
 ComparisonResult changes = comparer.detectChanges(
     previousScreen,
     currentScreen
 );

 if (changes.hasSignificantChanges()) {
     processUIUpdate(changes.getChangedRegions());
 }

Element Extraction


 // Extract UI elements from classification
 ContourExtractor extractor = new ContourExtractor();
 Mat classified = classifier.classify(scene);

 List<Match> elements = extractor.extractElements(
     classified,
     targetClassIndex
 );

Validation


 // Validate match results
 SizeComparator validator = new SizeComparator();

 boolean validSize = validator.isWithinExpectedSize(
     foundElement,
     expectedSize,
     tolerance
 );

Algorithm Options

Comparison Sensitivity

High: Detect pixel-level changes
Medium: Ignore minor variations
Low: Major structural changes only

Contour Extraction

External only: Outer boundaries
Hierarchical: Nested contours
Simple: Compressed contours

Performance Considerations

Full image comparison is computationally expensive
Consider region-of-interest comparison for speed
Contour extraction benefits from binary input
Cache comparison results when possible

Since:

1.0

See Also:

Classes

Class

Description

ContourExtractor

Extracts and processes contours from classified image regions to generate Match objects.

ContourExtractor.Builder

Builder class for constructing ContourExtractor instances with required parameters.

ImageComparer

Compares images to determine similarity and find matches between patterns.

SizeComparator

Determines size relationships between Pattern objects for image comparison operations.

Package io.github.jspinak.brobot.analysis.compare