DStretch® is widely known within the study of rock art and is used across the field of archaeology to enhance the visibility of faded or difficult to see motifs. DStretch calculates enhancements based on all RGB values in an image and so often struggles to decorrelate local areas with high correlation, particularly if there is a wide distribution of RGB values such as in dense rock art scenes, orthomosaics, or 3D model textures. Local decorrelation stretch (L-DCS) automatically separates an image into overlapping windows, performing decorrelated stretching on each, and merging them back into a single image. The results retained greater fine details in motifs and improved significantly the visibility of “hidden” motifs in densely painted areas when compared to DStretch. The flexibility afforded by our approach and the consistency of results in different scenarios make L-DCS a robust approach for automated DCS on larger datasets without manual intervention. • A more robust method for enhancing rock art motifs in dense scenes. • L-DCS can handle large images more effectively. • L-DCS handles greater variation in pigment colours without the need for multiple filters.
Dyson et al. (Sun,) studied this question.