ZEISS Maximum Birefringence Projector

When performing mineral classification for automated petrographic light microscopy data, the mineral crystals are typically randomly oriented relative to the direction of polarization. Complete information can be found by acquiring crossed polarized data at multiple relative orientations of polarizer/analyzer pair to the sample. This data is not optimally presented to allow for a machine learning algorithm to construct an accurate classification algorithm as the color, intensity and texture information that encodes mineral information is spread across all the different polarization orientations, presented as channels of the image. A maximal birefringence representation of the data, where each pixel is assigned the maximum intensity birefringence value from all crossed-polarized angles, allows the machine learning algorithm to directly lock into characteristic mineral signals.

ZEISS Maximum Birefringence Projector takes multi-polarized automated petrography data and re-constitutes it into a maximum-birefringence representation, with a channel represented in Red-Green-Blue (RGB) color space, and one in Hue-Saturation-Value (HSV) color space. As a change in polarization representation potentially engenders an image shift, a polarization channel registration utility is included to minimize the risk of image blurring. The re-ordered data is then saved in a format which transmits all the sample metadata in a format fully readable by ZEN microscopy image analysis software. This format is perfect to then be taken forward into subsequent mineral classification and automated analysis.