Learn_opencv-python

Image Proceesion

• load image: cv2.imread(path, num)
  • path: the relative or absolute path of the image
  • num: 0->color, 1->gray, 2->unchanged
• show image: cv2.imshow(name, src)
  • name: the window name
  • src: the target image
• flip image: cv2.flip(src, dst)
  • src: the target image
  • dst: set 0-vertical, 1-horizontal, -1-verical+horizontal to rotate the image
• linear transformation: cv2.addWeighted

Adaptive Threshold

• global threshold: cv2.threshold(src, thres, max, method)
  • src: target image
  • thres: the threshold value
  • max: the value replace the threshold value
  • method: cv2.THRESH_BINARY, cv2.THRESH_BINARY_INV, cv2.THRESH_TRUNC, cv2.THRESH_TOZERO, cv2.THRESH_TOZERO_INV
• local theshold: cv2.adaptiveThreshold(src, dst, max, method, type, blocksize, constant)
  • src: target image(input)
  • dst: the output image
  • max: maximum the threshold will separate the image into 0 and max
  • method: ADAPTIVE_THRESH_MEAN_C or ADAPTIVE_THRESH_GAUSSIAN_C
  • type: THRESH_BINARY or THRESH_BINARY_INV
  • blocksize: use odd number to decide the threshold value
  • constant: after do the threshold, it may minus some constant value

Transform

• rows, cols, channel = img.shape (color); rows, cols = img.shape (gray)
  • rows -> height
  • columns -> width

Translation: cv2.warpAffine(src, H, size)

• src: the image which you want to transform
• H: transform matrix
• size: the size(rows, columns) after transform

rotation: cv2.getRotationMatrix2D((x, y), angle, scale)

• (x, y): the central of the rotation image
• angle(theta): the angle of the rotation
• scale: the size of the image zoomed

Perspective transformation

cv2.getPerspectiveTransform(position1, position2)

• position1: the origin position(4 points)
• position2: the position you want to wrap(4 points)

resize the image: cv2.resize(src, size)

• src: target image
• size: the new size of the image

Gaussian and Laplacian Pyramid

cv2.pyrDown(src, dst, size)

• src: the image which you want to do pyramid down
• dst: the result image
• size: the size of the dst

cv2.pyrUp(src, dst, size)

• src: the image which you want to do pyramid down
• dst: the result image
• size: the size of the dst

Add two images: cv2.add(src1, src2)

• src1, src2: the two pictures you want to add each other

Edge Detection

• sobel: cv2.Sobel(src, ddepth, x, y ...)
  • src: the image which you want to find the edge
  • ddepth: the depth of the image
  • x, y: the differential of x y
• laplacian: cv2.Laplacian(src, ddepth)
  • src: the image which you want to find the edge
  • ddepth: the depth of the image
• canny: cv2.Canny(src, threshold_low, threshold_high)
  • src: the image which you want to find the edge
  • threshold_low: the lower bound of the image
  • threshold_high: the upper bound of the image