This project was an early investigation as to how one might analyse the multi- and hyper-spectral response of plant leaves against their 3D shape, using photometric stereo to capture high-resolution relief maps of plane rosettes.
Matlab | Python | Raspberry Pi
How Did It Work?
The literature [1, 2, 3] shows that the multi-spectral response of leaf matter is affected by its orientation relative to the camera and light-source. Since photometric stereo is an active illumination reconstruction method, this could have an affect, we intrinsically vary the illumination direction, while the orientation of the plant relative to the camera. This preliminary investigation sought to find out if photometric stereo reconstructions, captured at a range of light frequencies, could be used to model the affect of orientation on the response of an active multi-spectral imaging device.
The following images show a series of surface-normal maps, captured under warm white light. The top row has been rectified for global curvature induced by the non-ideal light sources, whilst the bottom is a raw normal map sample. Each colour channel indicates the magnitude of the surface normal
The initial results were promising, suggesting that under warm-white light we could capture usable surface reconstructions. This work was continued by Gytis Bernotas as part of his Ph.D research.
 VRINDTS, E., DE BAERDEMAEKER, J., AND RAMON, H. Weed detection using canopy reflection. Precision Agriculture 3, 1 (2002), 63–80.
 BEHMANN, J., MAHLEIN, A. K., PAULUS, S., KUHLMANN, H., OERKE, E. C., AND PLÜMER, L. Generation and application of hyperspectral 3D plant models. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 8928 (2015), 117–130.
 ROSCHER, R., BEHMANN, J., MAHLEIN, A. K., DUPUIS, J., KUHLMANN, H., AND PLÜMER, L. Detection of Disease Symptoms on Hyperspectral 3D Plant Models. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences 3, June (2016), 89–96.