Fig. 3

Two step image processing protocol for automated entry in the database. In the first step the raw images (a) are processed to identify the region of the blue germination paper containing roots (column on the left). This is achieved by determining the Hue value (b) of the image from the RGB data. The Hue image is then easily segmented using the Otsu auto-threshold approach and the contour of the region is then cropped (c). In the second step, the raw images are processed to identify and read the QR code in the image. This is achieved by determining the minimum image from the initial image data (a) and the image resulting from a variance filter (d). The resulting image (e) is then thresholded and cropped before it is passed into the Zbar decoder (f)