#!/usr/bin/env python

from pathlib import Path

import imageio

from matplotlib.pyplot import show, subplots

# from spectral import rx

from numpy import empty, zeros_like, around, uint8

from argparse import ArgumentParser

def main():

"""

./hyperspectral.py fig10.png --xc 841 1235 1623 --yc 41 442 863 --xs 50 325 --ys 50 350

./hyperspectral.py fig12.png --xc 734 1081 1392 --yc 32 377 863 --xs 50 300 --ys 50 325

"""

p = ArgumentParser('read and analyse hyperspectral image files')

p.add_argument('fn', help='file to analyse')

p.add_argument(

'--xc', help='x0 x1 x-pixel coordinates to clip', nargs=3, type=int)

p.add_argument(

'--yc', help='y0 y1 y-pixel coordinates to clip', nargs=3, type=int)

p.add_argument('--xs', help='x0 x1 x-pixel coordinates to clip within each subimage (ROI)',

nargs=2, type=int, default=(None, None))

p.add_argument('--ys', help='y0 y1 x-pixel coordinates to clip within each subimage (ROI)',

nargs=2, type=int, default=(None, None))

p.add_argument(

'-t', '--thres', help='threshold above which detection is declared', type=float, default=100)

p.add_argument('-l', '--minmax', help='min max pixel values in colormap (for plotting only)',

nargs=2, type=int, default=(None, None))

p = p.parse_args()

img, full = quadstack(p.fn, p.xc, p.yc, p.xs, p.ys)

imgf = imgfilt(img, p.thres)

plots(img, imgf, p.minmax, full, p.thres)

show()

def quadstack(fn, xc, yc, xs, ys):

fn = Path(fn).expanduser()

imgr = rgba2gray(imageio.imread(fn)) # this has all four quadrants

# %% extract four quadrants from one image, knowing a priori the system geometry performance

"""

"""

img = {} # we don't know if they'll all be the same size even due to affine distortion

img['4278'] = imgr[yc[0]:yc[1], xc[0]:xc[1]]

img['5577'] = imgr[yc[1]:yc[2], xc[0]:xc[1]]

img['7320'] = imgr[yc[0]:yc[1], xc[1]:xc[2]]

img['6300'] = imgr[yc[1]:yc[2], xc[1]:xc[2]]

# %% extract subimage

ims = empty((4, ys[1]-ys[0], xs[1]-xs[0]), dtype=uint8)

for i, I in enumerate(img):

ims[i, ...] = img[I][ys[0]:ys[1], xs[0]:xs[1]]

return ims, img

def imgfilt(img, thres):

"""

trivial intensity threshold "filter"

"""

imgf = zeros_like(img, dtype=bool)

imgf[img > thres] = True

return imgf

def plots(img, imgproc, mm, full, thres):

fg, ax = subplots(2, 2)

ax = ax.ravel()

for a, I, f in zip(ax, img, full):

# ,vmin=mm[0],vmax=mm[1])

hi = a.imshow(I, cmap='gray', interpolation='none')

fg.colorbar(hi, ax=a)

a.set_title(f)

a.axis('off')

fg.suptitle('original image')

fg, ax = subplots(2, 2)

ax = ax.ravel()

for a, I, f in zip(ax, imgproc, full):

hi = a.imshow(I, cmap='gray', interpolation='none', vmin=0, vmax=1)

# fg.colorbar(hi,ax=a)

a.set_title(f)

a.axis('off')

fg.suptitle('threshold image >= {:.0f}'.format(thres))

def rgba2gray(rgba):

"""

discards transparancy, assumes single m x n x 4 image

"""

return around(rgba[..., :-1].dot([0.299, 0.587, 0.114])).astype(rgba.dtype)

if __name__ == '__main__':

main()