Article Info
Analysis of Wavelet-Based Features for Identifying Similarities in Turtle Scute Patterns
Mohd Erman Safawie Che Ibrahim, Wan Nural Jawahir Hj Wan Yussof, Muhammad Suzuri Hitam, Ezmahamrul Afreen Awalludin, Mohamad Fathullah Ruslan, Siti NurFarahim Shaharudin
Abstract
Turtle scute identification is vital for ecological and conservation research but traditional methods, relying on manual observation and image comparison, are time-consuming and error-prone, especially with varying scales and orientations of scute patterns. This study explores wavelet-based features for analyzing similarities in turtle scute patterns. Utilizing multiple wavelet families, including Coif1, Sym2, Db1, and Haar, a comprehensive analysis of scute patterns was conducted by extracting from two images. Features such as energy, variance, standard deviation, waveform length, and entropy are computed from wavelet decompositions to evaluate their effectiveness in capturing subtle differences and complexities in the patterns. The findings highlight Coif1 as the most effective wavelet family, demonstrating higher Euclidean distances and greater sensitivity to variations in scute patterns. Notably, the study reveals consistent feature values across rotations (0?, 90?, 180?, and 270?), underscoring the reliability of these wavelet families in maintaining pattern recognition accuracy under different orientations. These results contribute valuable insights for advancing turtle identification methods based on their distinctive scute patterns.
keyword
Image Analysis, Wavelet Families, Feature Extraction, Wavelet-Based Features, Turtle Scute Identification.
Area
Pattern Recognition