# Bluring/Smoothing example using a 1D Gaussian Kernel 
# We show how a 1D kernel is not the same as a 2D kernel,
# See the smoothing_separable.py example to show how to use separable 
# 1D kernels to emulate the 2D kernel application, but much faster.
#
#
# Jay Summet 2015
#
#Python 2.7, OpenCV 2.4.x
#

import cv2
import numpy as np


#Linux window threading setup code.
cv2.startWindowThread()
cv2.namedWindow("Difference")
cv2.namedWindow("Gaussian Blur")
cv2.namedWindow("Gaussian 1D Blur")

#Load source / input image as grayscale, also works on color images...
imgIn = cv2.imread("blueSky.jpg", cv2.IMREAD_GRAYSCALE)


# 1D Gausian Kernel using sigma of 1.7
kernel = cv2.getGaussianKernel( 11, 1.7 )

#If you transpose the kernel the 1D filter operation is oriented
# horizontally instead of vertically.
#kernel = np.transpose(kernel)

oneD = cv2.filter2D(imgIn, -1, kernel)

cv2.imshow("Gaussian 1D Blur", oneD)

gBlurImg = cv2.GaussianBlur(imgIn, (11,11), 1.7)
cv2.imshow("Gaussian Blur", gBlurImg)

#This technique (differencing two images and showing the result)
# is a good general purpose way to compare two very similar, but not 
# quite the same images to more easily identify the differences.

diff = oneD- gBlurImg
cv2.imshow("Difference", diff)
cv2.waitKey(0)