Figure 2

Processing steps of cotton boll images. A native camera raw image (A) is separated into the monochromatic R, G, and B channels, and the B channel is discarded. To obtain the cotton boll mask, the green channel is filtered and reduced in size (B), then further blurred with a Butterworth lowpass filter (C). The mask is then obtained by thresholding (D). In parallel, a normalized intensity is generated by dividing the green channel by the red channel on a pixel-by-pixel basis (E). Multiplication with the mask yields the final image of normalized green intensity from the cotton boll, separated from background (F). The background in (F) is indicated by the gray checkerboard pattern and is excluded from any image analysis steps.