Accessing and Modifying pixel values
Pixel value |
|
Accessing only blue pixel |
|
Modifying A Pixel |
img[100,100] = [255,255,255]
|
Better pixel accessing |
|
Better pixel modifying |
img.itemset((10,10,2),100)
|
Access image properties |
|
Total number of pixels |
|
Image datatype |
|
Getting ROI |
ball = img[280:340, 330:390]
|
Setting ROI |
img[273:333, 100:160] = ball
|
Split Channels |
b,g,r = cv2.split(img)
b = img[:,:,0]
|
Making Borders for Images |
cv2.copyMakeBorder(img1,10,10,10,10,cv2.BORDER_REPLICATE)
|
borderType |
cv2.BORDER_CONSTANT
cv2.BORDER_REFLECT
cv2.BORDER_REFLECT_101
cv2.BORDER_REPLICATE
cv2.BORDER_WRAP
|
Arithmetic Operations on Images
Image Addition (OPENCV) |
print cv2.add(x,y) # 250+10 = 260 => 255
|
Image Addition (Numpy) |
print x+y # 250+10 = 260 % 256 = 4
|
Image Alpha Blending |
dst = cv2.addWeighted(img1,0.7,img2,0.3,0)
|
Bitwise AND |
img1_bg = cv2.bitwise_and(roi,roi,mask = mask_inv)
|
Bitwise NOT |
mask_inv = cv2.bitwise_not(mask)
|
Morphological Transformations
Erosion |
erosion = cv2.erode(img,kernel,iterations = 1)
|
Dilation |
dilation = cv2.dilate(img,kernel,iterations = 1)
|
Opening |
opening = cv2.morphologyEx(img, cv2.MORPH_OPEN, kernel)
|
Closing |
closing = cv2.morphologyEx(img, cv2.MORPH_CLOSE, kernel)
|
Morphological Gradient |
gradient = cv2.morphologyEx(img, cv2.MORPH_GRADIENT, kernel) |
Top Hat |
tophat = cv2.morphologyEx(img, cv2.MORPH_TOPHAT, kernel) |
Black Hat |
blackhat = cv2.morphologyEx(img, cv2.MORPH_BLACKHAT, kernel) |
Create Structuring Elements |
cv2.getStructuringElement(cv2.MORPH_RECT,(5,5))
cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(5,5))
cv2.getStructuringElement(cv2.MORPH_CROSS,(5,5))
|
|
|
Performance Measurement and Improvement Techniques
Find # of clock-cycles |
e1 = cv2.getTickCount()
# your code execution
e2 = cv2.getTickCount()
time = (e2 - e1)/ cv2.getTickFrequency()
|
Find clock cycles per second |
|
Enable Optimizations |
cv2.setUseOptimized(True)
|
Measure Performance (IPython) |
|
Performance Optimization Techniques |
1. Avoid using loops in Python as far as possible, especially double/triple loops etc. They are inherently slow.
2. Vectorize the algorithm/code to the maximum possible extent because Numpy and OpenCV are optimized for vector operations.
3. Exploit the cache coherence.
4. Never make copies of array unless it is needed. Try to use views instead. Array copying is a costly operation. |
Geometric Transformations of Images
Scaling Types |
cv2.INTER_AREA
cv2.INTER_CUBIC
cv2.INTER_LINEAR
|
Scaling |
res = cv2.resize(img,(2width, 2height), interpolation = cv2.INTER_CUBIC)
|
Shifting (100 x 50) |
M = np.float32([[1,0,100],[0,1,50]])
dst = cv2.warpAffine(img,M,(cols,rows))
|
Rotation |
M = cv2.getRotationMatrix2D((cols/2,rows/2),90,1)
dst = cv2.warpAffine(img,M,(cols,rows))
|
Affine Transformation |
pts1 = np.float32([[50,50],[200,50],[50,200]])
pts2 = np.float32([[10,100],[200,50],[100,250]])
M = cv2.getAffineTransform(pts1,pts2)
dst = cv2.warpAffine(img,M,(cols,rows))
|
Perspective Transformation |
pts1 = np.float32([[56,65],[368,52],[28,387],[389,390]])
pts2 = np.float32([[0,0],[300,0],[0,300],[300,300]])
M = cv2.getPerspectiveTransform(pts1,pts2)
dst = cv2.warpPerspective(img,M,(300,300))
|
Canny Edge Detection
Canny Detection |
edges = cv2.Canny(img,100,200)
|
Image Pyramids
Lower Gaussian Pyramid |
lower_reso = cv2.pyrDown(higher_reso)
|
Higher Gaussian Pyramid |
higher_reso2 = cv2.pyrUp(lower_reso)
|
Pyramid Blending |
1. Load the two images
2. Find the Gaussian Pyramids
3. From Gaussian Pyramids, find their Laplacian Pyramids
4. Now each levels of Laplacian Pyramids
5. Finally from this joint image pyramids, reconstruct the original image |
|
|
Changing Colorspaces
List Colorspace Flags (150+) |
flags = [i for i in dir(cv2) if i.startswith('COLOR_')
|
Convert to Gray |
img_gray = cv2.cvtColor(img, cv2. COLOR_BGR2GRAY)
|
Convert to hsv |
hsv = cv2.cvtColor(img, cv2. COLOR_BGR2HSV)
|
Track Blue (color) Object |
lower_blue = np.array([110,50,50])
upper_blue = np.array([130,255,255])
mask = cv2.inRange(hsv, lower_blue, upper_blue)
res = cv2.bitwise_and(frame,frame, mask= mask)
|
Find HSV Color |
green = np.uint8([[[0,255,0 ]]])
hsv_green = cv2.cvtColor(green,cv2.COLOR_BGR2HSV)
|
Image Thresholding
Thresholding Types |
cv2.THRESH_BINARY
cv2.THRESH_BINARY_INV
cv2.THRESH_TRUNC
cv2.THRESH_TOZERO
cv2.THRESH_TOZERO_INV
|
Getting Threshold |
ret,thresh4 = cv2.threshold(img,127,255,cv2.THRESH_TOZERO)
|
Adaptive Method Types |
cv2.ADAPTIVE_THRESH_MEAN_C
cv2.ADAPTIVE_THRESH_GAUSSIAN_C
|
Adaptive Threshold |
th3 = cv2.adaptiveThreshold(img,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY,11,2)
|
Otsu’s Binarization |
ret3,th3 = cv2.threshold(blur,0,255,cv2.THRESH_BINARY+cv2.THRESH_OTSU)
|
Smoothing Images
Convolve an Image |
dst = cv2.filter2D(img,-1,kernel)
|
Box (averaging) Filtering |
blur = cv2.blur(img,(5,5))
cv2.boxFilter()
|
Create Gaussian Kernel |
cv2.getGaussianKernel(size, sigma, type)
|
Gaussian Blur |
blur = cv2.GaussianBlur(img,(5,5),0)
|
Median Blur |
median = cv2.medianBlur(img,5)
|
Bilateral Blur |
blur = cv2.bilateralFilter(img,9,75,75)
|
Image Gradients
Sobel |
sobelx = cv2.Sobel(img,cv2.CV_64F,1,0,ksize=5)
|
Laplacian |
laplacian = cv2.Laplacian(img,cv2.CV_64F)
|
*Output datatype cv2.CV_8U or np.uint8. So when you convert data to np.uint8, all negative slopes are made zero. In simple words, you miss that edge. If you want to detect both edges, better option is to keep the output datatype to some higher forms, like cv2.CV_16S, cv2.CV_64F etc, take its absolute value and then convert back to cv2.CV_8U
|
