def grab_cut_mask(img_col, mask, debug=False):
assert isinstance(img_col, numpy.ndarray), 'image must be a numpy array'
assert isinstance(mask, numpy.ndarray), 'mask must be a numpy array'
assert img_col.ndim == 3, 'skin detection can only work on color images'
assert mask.ndim == 2, 'mask must be 2D'
kernel = numpy.ones((50, 50), numpy.float32) / (50 * 50)
dst = cv2.filter2D(mask, -1, kernel)
dst[dst != 0] = 255
free = numpy.array(cv2.bitwise_not(dst), dtype=numpy.uint8)
if debug:
scripts.display('not skin', free)
scripts.display('grabcut input', mask)
grab_mask = numpy.zeros(mask.shape, dtype=numpy.uint8)
grab_mask[:, :] = 2
grab_mask[mask == 255] = 1
grab_mask[free == 255] = 0
if numpy.unique(grab_mask).tolist() == [0, 1]:
logger.debug('conducting grabcut')
bgdModel = numpy.zeros((1, 65), numpy.float64)
fgdModel = numpy.zeros((1, 65), numpy.float64)
if img_col.size != 0:
mask, bgdModel, fgdModel = cv2.grabCut(img_col, grab_mask, None,
bgdModel, fgdModel, 5,
cv2.GC_INIT_WITH_MASK)
mask = numpy.where((mask == 2) | (mask == 0), 0,
1).astype(numpy.uint8)
else:
logger.warning('img_col is empty')
return mask
def get_rgb_mask(img, debug=False):
assert isinstance(img, numpy.ndarray), 'image must be a numpy array'
assert img.ndim == 3, 'skin detection can only work on color images'
logger.debug('getting rgb mask')
lower_thresh = numpy.array([45, 52, 108], dtype=numpy.uint8)
upper_thresh = numpy.array([255, 255, 255], dtype=numpy.uint8)
mask_a = cv2.inRange(img, lower_thresh, upper_thresh)
mask_b = 255 * ((img[:, :, 2] - img[:, :, 1]) / 20)
mask_c = 255 * ((numpy.max(img, axis=2) - numpy.min(img, axis=2)) / 20)
mask_a = mask_a.astype(float)
print(mask_a.dtype)
print(mask_b.dtype)
print(mask_c.dtype)
msk_rgb = cv2.bitwise_and(mask_a, mask_b)
msk_rgb = cv2.bitwise_and(mask_c, msk_rgb)
msk_rgb[msk_rgb < 128] = 0
msk_rgb[msk_rgb >= 128] = 255
if debug:
scripts.display('input', img)
scripts.display('mask_rgb', msk_rgb)
return msk_rgb.astype(float)
def main(args):
assert isinstance(args, argparse.Namespace), 'args must be of type argparse.Namespace not {0}'.format(type(args))
sift = cv2.SIFT()
flann = cv2.FlannBasedMatcher({'algorithm': 0, 'trees': 5}, {'checks': 50})
result, result_gry = None, None
for path in args.paths:
try:
assert os.path.exists(path), '{0} is not a valid path'.format(path)
logger.info('processing {0}'.format(path))
cam = cv2.VideoCapture(path)
logger.debug('opened video')
count = 0
while True:
logger.debug('reading frame {0}'.format(count))
ret, frame = cam.read()
logger.debug('cropping frame to minimise distortions')
#todo: Crop frame!
if ret:
logger.debug('frame read correctly')
count += 1
if result is not None and result_gry is not None:
frame_gry = cv2.cvtColor(frame, cv2.COLOR_RGB2GRAY)
logger.debug('computing sift features')
keypoints0, descriptors0 = sift.detectAndCompute(result_gry, None)
keypoints1, descriptors1 = sift.detectAndCompute(frame_gry, None)
logger.debug('finding correspondence')
matches = flann.knnMatch(descriptors0, descriptors1, k=args.knn)
positive = []
for match0, match1 in matches:
if match0.distance < args.lowe*match1.distance:
positive.append(match0)
if len(positive) > args.min_correspondence:
src_pts = numpy.array([keypoints0[good_match.queryIdx].pt for good_match in positive], dtype=numpy.float32)
src_pts = src_pts.reshape((-1, 1, 2))
dst_pts = numpy.array([keypoints1[good_match.trainIdx].pt for good_match in positive], dtype=numpy.float32)
dst_pts = dst_pts.reshape((-1, 1, 2))
M, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC, 5.0)
result = combine.combine_images(frame, result, M)
if args.display and not args.quiet:
scripts.display('result', result)
if cv2.waitKey(25) & 0xFF == ord('q'):
break
else:
logger.warning('too few correspondence points')
else:
result = frame
result_gry = cv2.cvtColor(result, cv2.COLOR_RGB2GRAY)
else:
break
logger.debug('{0} is completed'.format(path))
cam.release()
cv2.destroyAllWindows()
if args.save:
scripts.save_image(path, result)
except Exception as error:
logger.warning('Failed to process {0}'.format(path))
logger.debug('Error msg: {0}'.format(error))
return result
def get_mask_ycrcb(self, img):
self.assert_image(img)
lower_thresh = numpy.array([90, 100, 130], dtype=numpy.uint8)
upper_thresh = numpy.array([230, 120, 180], dtype=numpy.uint8)
img_ycrcb = cv2.cvtColor(img, cv2.COLOR_RGB2YCR_CB)
msk_ycrcb = cv2.inRange(img_ycrcb, lower_thresh, upper_thresh)
if self.args.debug:
scripts.display('input', img)
scripts.display('mask_ycrcb', msk_ycrcb)
self.add_mask(msk_ycrcb)
def get_mask_ycrcb(self, img):
self.assert_image(img)
lower_thresh = numpy.array([90, 100, 130], dtype=numpy.uint8)
upper_thresh = numpy.array([230, 120, 180], dtype=numpy.uint8)
img_ycrcb = cv2.cvtColor(img, cv2.COLOR_RGB2YCR_CB)
msk_ycrcb = cv2.inRange(img_ycrcb, lower_thresh, upper_thresh)
if self.args.debug:
scripts.display("input", img)
scripts.display("mask_ycrcb", msk_ycrcb)
self.add_mask(msk_ycrcb)
def get_mask_hsv(self, img):
logger.debug('Applying hsv threshold')
self.assert_image(img)
lower_thresh = numpy.array([0, 50, 0], dtype=numpy.uint8)
upper_thresh = numpy.array([120, 150, 255], dtype=numpy.uint8)
img_hsv = cv2.cvtColor(img, cv2.COLOR_RGB2HSV)
msk_hsv = cv2.inRange(img_hsv, lower_thresh, upper_thresh)
if self.args.debug:
scripts.display('input', img)
scripts.display('mask_hsv', msk_hsv)
self.add_mask(msk_hsv)
def get_mask_hsv(self, img):
logger.debug("Applying hsv threshold")
self.assert_image(img)
lower_thresh = numpy.array([0, 50, 0], dtype=numpy.uint8)
upper_thresh = numpy.array([120, 150, 255], dtype=numpy.uint8)
img_hsv = cv2.cvtColor(img, cv2.COLOR_RGB2HSV)
msk_hsv = cv2.inRange(img_hsv, lower_thresh, upper_thresh)
if self.args.debug:
scripts.display("input", img)
scripts.display("mask_hsv", msk_hsv)
self.add_mask(msk_hsv)
def get_mask_rgb(self, img):
logger.debug('Applying rgb thresholds')
lower_thresh = numpy.array([45, 52, 108], dtype=numpy.uint8)
upper_thresh = numpy.array([255, 255, 255], dtype=numpy.uint8)
mask_a = cv2.inRange(img, lower_thresh, upper_thresh)
mask_b = 255*((img[:, :, 2]-img[:, :, 1])/20)
logger.debug('mask_b unique: {0}'.format(numpy.unique(mask_b)))
mask_c = 255*((numpy.max(img, axis=2)-numpy.min(img, axis=2))/20)
logger.debug('mask_d unique: {0}'.format(numpy.unique(mask_c)))
msk_rgb = cv2.bitwise_and(mask_a, mask_b)
msk_rgb = cv2.bitwise_and(mask_c, msk_rgb)
if self.args.debug:
scripts.display('input', img)
scripts.display('mask_rgb', msk_rgb)
self.add_mask(msk_rgb)
def get_hsv_mask(img, debug=False):
assert isinstance(img, numpy.ndarray), 'image must be a numpy array'
assert img.ndim == 3, 'skin detection can only work on color images'
logger.debug('getting hsv mask')
lower_thresh = numpy.array([0, 50, 0], dtype=numpy.uint8)
upper_thresh = numpy.array([120, 150, 255], dtype=numpy.uint8)
img_hsv = cv2.cvtColor(img, cv2.COLOR_RGB2HSV)
msk_hsv = cv2.inRange(img_hsv, lower_thresh, upper_thresh)
msk_hsv[msk_hsv < 128] = 0
msk_hsv[msk_hsv >= 128] = 1
if debug:
scripts.display('input', img)
scripts.display('mask_hsv', msk_hsv)
return msk_hsv.astype(float)
def get_ycrcb_mask(img, debug=False):
assert isinstance(img, numpy.ndarray), 'image must be a numpy array'
assert img.ndim == 3, 'skin detection can only work on color images'
logger.debug('getting ycrcb mask')
lower_thresh = numpy.array([90, 100, 130], dtype=numpy.uint8)
upper_thresh = numpy.array([230, 120, 180], dtype=numpy.uint8)
img_ycrcb = cv2.cvtColor(img, cv2.COLOR_RGB2YCR_CB)
msk_ycrcb = cv2.inRange(img_ycrcb, lower_thresh, upper_thresh)
msk_ycrcb[msk_ycrcb < 128] = 0
msk_ycrcb[msk_ycrcb >= 128] = 1
if debug:
scripts.display('input', img)
scripts.display('mask_ycrcb', msk_ycrcb)
return msk_ycrcb.astype(float)
def evaluate(img_col, args):
numpy.seterr(all='ignore')
assert isinstance(img_col, numpy.ndarray), 'img_col must be a numpy array'
assert img_col.ndim == 3, 'img_col must be a color image ({0} dimensions currently)'.format(img_col.ndim)
assert isinstance(args, argparse.Namespace), 'args must be of type argparse.Namespace not {0}'.format(type(args))
img_gry = cv2.cvtColor(img_col, cv2.COLOR_RGB2GRAY)
rows, cols = img_gry.shape
crow, ccol = rows/2, cols/2
f = numpy.fft.fft2(img_gry)
fshift = numpy.fft.fftshift(f)
fshift[crow-75:crow+75, ccol-75:ccol+75] = 0
f_ishift = numpy.fft.ifftshift(fshift)
img_fft = numpy.fft.ifft2(f_ishift)
img_fft = 20*numpy.log(numpy.abs(img_fft))
if args.display and not args.testing:
cv2.destroyAllWindows()
scripts.display('img_fft', img_fft)
scripts.display('img_col', img_col)
cv2.waitKey(0)
result = numpy.mean(img_fft)
return img_fft, result, result < args.thresh
def evaluate(img_col, args):
numpy.seterr(all='ignore')
assert isinstance(img_col, numpy.ndarray), 'img_col must be a numpy array'
assert img_col.ndim == 3, 'img_col must be a color image ({0} dimensions currently)'.format(img_col.ndim)
assert isinstance(args, argparse.Namespace), 'args must be of type argparse.Namespace not {0}'.format(type(args))
img_gry = cv2.cvtColor(img_col, cv2.COLOR_RGB2GRAY)
rows, cols = img_gry.shape
crow, ccol = rows//2, cols//2
f = numpy.fft.fft2(img_gry)
fshift = numpy.fft.fftshift(f)
fshift[crow-75:crow+75, ccol-75:ccol+75] = 0
f_ishift = numpy.fft.ifftshift(fshift)
img_fft = numpy.fft.ifft2(f_ishift)
img_fft = 20*numpy.log(numpy.abs(img_fft))
if args.display and not args.testing:
cv2.destroyAllWindows()
scripts.display('img_fft', img_fft)
scripts.display('img_col', img_col)
cv2.waitKey(0)
result = numpy.mean(img_fft)
return img_fft, result, result < args.thresh
def evaluate(img_col, args):
numpy.seterr(all='ignore')
assert isinstance(img_col, numpy.ndarray), 'img_col must be a numpy array'
assert img_col.ndim == 3, 'img_col must be a color image ({0} dimensions currently)'.format(img_col.ndim)
assert isinstance(args, argparse.Namespace), 'args must be of type argparse.Namespace not {0}'.format(type(args))
img_gry = cv2.cvtColor(img_col, cv2.COLOR_RGB2GRAY)
rows, cols = img_gry.shape
crow, ccol = rows/2, cols/2
cv2.imwrite("test.png",img_gry)
data = Image.open("test.png")
remmax = lambda x: x/x.max()
remmin = lambda x: x - numpy.amin(x, axis=(0,1), keepdims=True)
touint8 = lambda x: (remmax(remmin(x))*(256-1e-4)).astype(int)
channels = data.split()
result_array = numpy.zeros_like(data)
f = numpy.fft.fft2(channels[0])
fshift = numpy.fft.fftshift(f)
fshift[int(crow-1):int(crow+1), int(ccol-1):int(ccol+1)] = 0
result_array[...]=touint8(fshift)
result_image = Image.fromarray(result_array)
result_image.save('out.png')
print(numpy.count_nonzero(result_array > 128 ))
fshift[int(crow-75):int(crow+75), int(ccol-75):int(ccol+75)] = 0
f_ishift = numpy.fft.ifftshift(fshift)
img_fft = numpy.fft.ifft2(f_ishift)
img_fft = 20*numpy.log(numpy.abs(img_fft))
if args.display and not args.testing:
cv2.destroyAllWindows()
scripts.display('img_fft', img_fft)
scripts.display('img_col', img_col)
cv2.waitKey(0)
result = numpy.mean(img_fft)
print (result)
return img_fft, result, result < args.thresh
else:
return evaluate(img_col=img_col, args=args)
if __name__ == '__main__':
args = scripts.get_args()
logger = scripts.get_logger(quite=args.quite, debug=args.debug)
x_okay, y_okay = [], []
x_blur, y_blur = [], []
for path in args.image_paths:
for img_path in scripts.find_images(path):
logger.debug('evaluating {0}'.format(img_path))
img = cv2.imread(img_path)
if isinstance(img, numpy.ndarray):
if args.testing:
scripts.display('dialog (blurry: Y?)', img)
blurry = False
if cv2.waitKey(0) in map(lambda i: ord(i), ['Y', 'y']):
blurry = True
x_axis = [1, 3, 5, 7, 9]
for x in x_axis:
img_mod = cv2.GaussianBlur(img, (x, x), 0)
y = evaluate(img_mod, args=args)[0]
if blurry:
x_blur.append(x)
y_blur.append(y)
else:
x_okay.append(x)
y_okay.append(y)
elif args.mask:
msk, res, blurry = FocusMask.blur_mask(img)
dst_pts = dst_pts.reshape((-1, 1, 2))
M, mask = cv2.findHomography(src_pts, dst_pts, cv2.RANSAC, 5.0)
result = combine.combine_images(frame, result, M)
if args.display and not args.quiet:
scripts.display('result', result)
if cv2.waitKey(25) & 0xFF == ord('q'):
break
else:
logger.warning('too few correspondence points')
else:
result = frame
result_gry = cv2.cvtColor(result, cv2.COLOR_RGB2GRAY)
else:
break
logger.debug('{0} is completed'.format(path))
cam.release()
cv2.destroyAllWindows()
if args.save:
scripts.save_image(path, result)
except Exception as error:
logger.warning('Failed to process {0}'.format(path))
logger.debug('Error msg: {0}'.format(error))
return result
if __name__ == '__main__':
args = scripts.get_args()
logger = scripts.get_logger(quiet=args.quiet, debug=args.debug)
result = main(args)
scripts.display('final result', result)
cv2.waitKey(0)
msk[:, w - dw:] = 255
return msk
def blur_mask(img):
assert isinstance(img, numpy.ndarray), 'img_col must be a numpy array'
assert img.ndim == 3, 'img_col must be a color image ({0} dimensions currently)'.format(
img.ndim)
msk, val, blurry = main.blur_detector(img)
logger.debug('inverting img_fft')
msk = cv2.convertScaleAbs(255 - (255 * msk / numpy.max(msk)))
msk[msk < 50] = 0
msk[msk > 127] = 255
logger.debug('removing border')
msk = remove_border(msk)
logger.debug('applying erosion and dilation operators')
msk = morphology(msk)
logger.debug('evaluation complete')
result = numpy.sum(msk) / (255.0 * msk.size)
logger.info('{0}% of input image is blurry'.format(int(100 * result)))
return msk, result, blurry
if __name__ == '__main__':
img_path = raw_input("Please Enter Image Path: ")
img = cv2.imread(img_path)
msk, val = blur_mask(img)
scripts.display('img', img)
scripts.display('msk', msk)
cv2.waitKey(0)
else:
return evaluate(img_col=img_col, args=args)
if __name__ == '__main__':
args = scripts.get_args()
logger = scripts.get_logger(quite=args.quite, debug=args.debug)
x_okay, y_okay = [], []
x_blur, y_blur = [], []
for path in args.image_paths:
for img_path in scripts.find_images(path):
logger.debug('evaluating {0}'.format(img_path))
img = cv2.imread(img_path)
if isinstance(img, numpy.ndarray):
if args.testing:
scripts.display('dialog (blurry: Y?)', img)
blurry = False
if cv2.waitKey(0) in map(lambda i: ord(i), ['Y', 'y']):
blurry = True
x_axis = [1, 3, 5, 7, 9]
for x in x_axis:
img_mod = cv2.GaussianBlur(img, (x, x), 0)
y = evaluate(img_mod, args=args)[0]
if blurry:
x_blur.append(x)
y_blur.append(y)
else:
x_okay.append(x)
y_okay.append(y)
elif args.mask:
msk, res, blurry = FocusMask.blur_mask(img)
msk[:, :dw] = 255
msk[:, w-dw:] = 255
return msk
def blur_mask(img):
assert isinstance(img, numpy.ndarray), 'img_col must be a numpy array'
assert img.ndim == 3, 'img_col must be a color image ({0} dimensions currently)'.format(img.ndim)
msk, val, blurry = main.blur_detector(img)
logger.debug('inverting img_fft')
msk = cv2.convertScaleAbs(255-(255*msk/numpy.max(msk)))
msk[msk < 50] = 0
msk[msk > 127] = 255
logger.debug('removing border')
msk = remove_border(msk)
logger.debug('applying erosion and dilation operators')
msk = morphology(msk)
logger.debug('evaluation complete')
result = numpy.sum(msk)/(255.0*msk.size)
logger.info('{0}% of input image is blurry'.format(int(100*result)))
return msk, result, blurry
if __name__ == '__main__':
img_path = raw_input("Please Enter Image Path: ")
img = cv2.imread(img_path)
msk, val = blur_mask(img)
scripts.display('img', img)
scripts.display('msk', msk)
cv2.waitKey(0)
#!/usr/bin/env python
# -*- coding: utf-8 -*-
__author__ = 'Will Brennan'
# Built-in Modules
import logging
# Standard Modules
import cv2
import numpy
# Custom Modules
import main
import scripts
if __name__ == '__main__':
args = scripts.get_args(from_file=False)
logger = scripts.get_logger(quite=args.quite, debug=args.debug)
cam = cv2.VideoCapture(0)
while True:
ret, img_col = cam.read()
img_msk = main.process(img_col, args=args)
if not args.display:
scripts.display('img_col', img_col)
scripts.display('img_msk', img_msk)
scripts.display('img_skn', cv2.bitwise_and(img_col, img_col, mask=img_msk))
cv2.waitKey(5)
#!/usr/bin/env python
# -*- coding: utf-8 -*-
__author__ = 'Will Brennan'
# Built-in Modules
import logging
# Standard Modules
import cv2
import numpy
# Custom Modules
import main
import scripts
if __name__ == '__main__':
args = scripts.get_args(from_file=False)
logger = scripts.get_logger(quite=args.quite, debug=args.debug)
cam = cv2.VideoCapture(0)
while True:
ret, img_col = cam.read()
img_msk = main.process(img_col, args=args)
if not args.display:
scripts.display('img_col', img_col)
scripts.display('img_msk', img_msk)
scripts.display('img_skn',
cv2.bitwise_and(img_col, img_col, mask=img_msk))
cv2.waitKey(5)
