Feature Extraction for Infrared Images

This code was independently developed by the author for infrared image feature extraction. The implementation calculates seven discriminative features from infrared imagery: aspect ratio (bounding box dimension analysis), complexity (contour-based perimeter-to-area measurements), compactness (shape regularity quantification), mean ratio (intensity distribution analysis), and invariant moments (rotation/scale-translation resistant Hu moments). These characteristics serve as critical preprocessing for recognition pipelines, with algorithmic optimizations enhancing computational efficiency and extraction accuracy. The feature vectors are structured to support machine learning classifiers through dimensionality-reduced representations. Additional performance improvements include optimized matrix operations for moment calculations and boundary detection enhancements for robust contour extraction. This implementation aims to support infrared-based research applications requiring reliable feature descriptors.