def multiplyImages(self):
print('Multiply color image {} with image {}'.format(
self.firstDecoder.name, self.secondDecoder.name))
unification = Unification(self.firstDecoder.name,
self.secondDecoder.name, self.imageType)
firstImage, secondImage = unification.colorUnification()
width, height = firstImage.shape[0], firstImage.shape[1]
maxValue = float(numpy.iinfo(firstImage.dtype).max)
result = numpy.ones((height, width, 3), numpy.uint8)
for h in range(height):
for w in range(width):
result[h, w, 0] = int(firstImage[h, w, 0]) * int(
secondImage[h, w, 0]) / maxValue
result[h, w, 1] = int(firstImage[h, w, 1]) * int(
secondImage[h, w, 1]) / maxValue
result[h, w, 2] = int(firstImage[h, w, 2]) * int(
secondImage[h, w, 2]) / maxValue
ImageHelper.Save(result, self.imageType, 'multiply-color-images',
False, self.firstDecoder, self.secondDecoder)
result = Commons.Normalization(firstImage, result)
ImageHelper.Save(result, self.imageType, 'multiply-color-images', True,
self.firstDecoder, self.secondDecoder)
def blendImages(self, ratio):
print('Blending gray image {} with image {} and ratio {}'.format(
self.firstDecoder.name, self.secondDecoder.name, ratio))
if ratio < 0 or ratio > 1.0:
raise ValueError('ratio is wrong')
unification = Unification(self.firstDecoder.name,
self.secondDecoder.name, 'L')
firstImage = unification.scaleUpGray(self.firstDecoder)
secondImage = unification.scaleUpGray(self.secondDecoder)
width, height = firstImage.shape[0], firstImage.shape[1]
result = numpy.ones((height, width), numpy.uint8)
for h in range(height):
for w in range(width):
pom = ratio * firstImage[h, w] + (1 - ratio) * secondImage[h,
w]
result[h, w] = pom
ImageHelper.Save(result, self.imageType, 'blend-gray-images', False,
self.firstDecoder, None, ratio)
def logarithm(self):
print('Logarithm color image {}'.format(self.firstDecoder.name))
height, width = self.firstDecoder.height, self.firstDecoder.width
image = self.firstDecoder.getPixels()
maxValue = float(numpy.iinfo(image.dtype).max)
maxR = numpy.amax(image[:, :, 0])
maxG = numpy.amax(image[:, :, 1])
maxB = numpy.amax(image[:, :, 2])
result = numpy.ones((height, width, 3), numpy.uint32)
for h in range(height):
for w in range(width):
result[h, w, 0] = maxValue * (math.log10(1 + image[h, w, 0]) /
math.log10(1 + maxR))
result[h, w, 1] = maxValue * (math.log10(1 + image[h, w, 1]) /
math.log10(1 + maxG))
result[h, w, 2] = maxValue * (math.log10(1 + image[h, w, 2]) /
math.log10(1 + maxB))
ImageHelper.Save(result.astype(numpy.uint8), self.imageType,
'logarithm-color', False, self.firstDecoder)
def getAdaptiveThreshold(self, maxVal, blockSize, c):
outputAdaptive = cv2.adaptiveThreshold(self.image, maxVal,
cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
cv2.THRESH_BINARY, blockSize, c)
ImageHelper.showImage('Adaptive Thresh', outputAdaptive)
def preProcessGrayScale(self):
cv2.equalizeHist(self.image, self.image)
ImageHelper.showImage('Normalized', self.image)
pass
def doStuff(self, fileName):
pass
hsvImg = cv2.cvtColor(self.image, cv2.COLOR_RGB2HSV)
h, s, v = cv2.split(hsvImg)
if FeatureDebug.DEBUG_DRAW_TRUTH:
from db import Database as db
self.db = db
self.image = self.__drawTruth__(self.image, fileName)
processedH = self.image.copy()
processedS = self.image.copy()
_, h_thresh = cv2.threshold(h, 60, 255, cv2.THRESH_BINARY_INV)
_, s_threshed = cv2.threshold(s, 50, 60, cv2.THRESH_BINARY)
#apply mask to image
processedH = cv2.bitwise_and(self.image, self.image, mask=h_thresh)
processedS = cv2.bitwise_and(self.image, self.image, mask=s_threshed)
ImageHelper.showImage('Original', self.image)
# ImageHelper.showImage('HSV', hsvImg)
ImageHelper.showImage('H', h)
ImageHelper.showImage('S', s)
ImageHelper.showImage('V', v)
ImageHelper.showImage('h_threshed', h_thresh)
# ImageHelper.showImage('h_mask_threshed', self.__drawTruth__(processedH, fileName))
ImageHelper.showImage('s_threshed', s_threshed)
# ImageHelper.showImage('s_mask_threshed', self.__drawTruth__(processedS, fileName))
return processedH