MULTISPECTRAL PICTOGRAPHS

Pictographs are an ancient art, which are painted on the rocks with pigments instead of being carved, and hence are much more sensitive to weathering and erosion. Images of pictographs are often need to be enhanced to make the art recognizable. One of the ways to do it is to use DStretch plug in for ImageJ written by Jon Harman. The software uses decorrelation stretch algorithms on single picture taken in visible spectrum with regular camera to improve the visibility of colors.

During my visit to Anza-Borrego Desert State Park I had a chance to take pictures of several pictographs along the Anza-Borrego Pictograph Trail. I used full-spectrum modified Sony A100 DSLR camera, Novoflex Noflexar 35mm F/3.5 lens and three filters: X-NiteCC1 filter from LDP LLC for visible spectrum, Baader U2 filter for near ultraviolet, and Hoya R072 filter for near infrared.

These pictographs are painted with red and yellow pigments on the granite surface and are still quite distinct, in spite of many ears of weathering. Nonetheless, the red pigment is more prominent and better preserved, while yellow pigment is fading and obscure. Processing this picture with DStretch gives some remarkable results, of which I only show two examples. DStretch allows to enhance the red pigment and separate it from the rest, using decorrelation stretch in YRE color space (above), while other algorithms (for example YWE color space) will also make yellow pigments more distinct.

Pictures of pictographs taken outside the visible spectrum gave very different results. Pictograph pattern is still visible in reflected ultraviolet, although it is impossible to discern different colors. On the other hand, picture shot in reflected infrared does not show any trace of pigment at all. Therefore, it might be possible to use IR-picture to subtract distracting granite pattern from the visible image in post-processing to make pigments more visible.

But, instead of doing it manually in image-editing software, I used "The Mirror's Surface Breaks" (TMSB) written by Ben Lincoln. It also uses decorrelation stretch as part of its workflow. Three images (visible, near-ultraviolet and near-infrared) were processed in TMSB using "Photography - Moderate - 01" processing configuration. Out of 255 generated images, only four were selected.

Many more pictures clearly enhanced both red and yellow pigments, albeit expressed as false colors, but these four were also quite pleasing aesthetically to me. More tests and experiments are needed to decide whether inclusion of ultraviolet and infrared images improves the separation between pigments and underlying rock surface. The results will also be different for different pigments and rock or stone substrate.