Fig. 2From: rosettR: protocol and software for seedling area and growth analysisOverview of image analysis steps in rosettR. Step 1: eccentricity of the plate is corrected for using an optimization algorithm that ensures that as many dark pixels as possible are within a circle matching the size of the plate. Rotation is corrected for by searching for clear peaks in the marginal distributions when step-wise rotating the image. Step 2: establishing a suitable threshold for background is done by a 2-component mixture model classifying pixels to the bright (backlit) background or all other darker pixels as plant material. Step 3: detected image features are sorted to the square in the grid they occupy the most. Step 4: a quality control image is generated that indicates the correction measures and colors the image features to indicate if they were detected correctlyBack to article page