Fig. 1. Summary of high-throughput, multi-faceted imaging methods.

The workflow runs from top to bottom, and the main features are highlighted. a An exemplar image showing the high throughput capacity of our imaging system. b Multispectral images (upper two rows) can be summarized as false color images (lower row) by principal component analysis, where red corresponds to PC1, green to PC2, and blue to PC3. c Full wing shape can be virtually reconstructed using information from dorsal and ventral segmentation. d A universal coordinate system for each wing can be generated automatically based on four landmarks (labeled as red dots). e Summarized wing shapes of two groups of butterflies: Lycaenidae on the left-hand side and Papilionidae and Nymphalidae on the right-hand side. Tail probabilities (Tail Prob.), curviness (Curv.), and the standard error of tail curvature (S.E. of Tail Curv.) are color coded accordingly. f Body and antennal morphologies can be measured automatically during image processing. g The summary of reflectance of four exemplar spectral bands (RGB and UV) from three specimens are shown. h Variation in UV reflectance of a group of butterflies is summarized as ‘UV signal,’ which represents the average contrasts of UV reflectance. Blue indicates low signal, and red indicates high.