Computer-vision

OpenCV(OpenSourceComputerVision),across-platformandfreetouselibraryoffunctionsisbasedonreal-timeComputerVisionwhichsupportsDeepLearningframeworksthataidsinimageandvideoprocessing.InComputerVision,theprincipalelementistoextractthepixelsfromtheimagetostudytheobjectsandthusunderstandwhatitcontains.BelowareafewkeyaspectsthatComputerVisionseekstorecognizeinthephotographs

NUMPYLIBRARY

Numpyis a very popular library for easily creating single, multidimensional array and matrices. It has a large collection of the mathematical function for performing an operation on these arrays.

#creat array
import numpy as np 
arr=np.array([[1,2,3],[4,5,6]])

OPENCV LIBRARY

OpenCVisahugeopen-sourcelibraryforcomputervision,machinelearning,andimageprocessing.OpenCVsupportsawidevarietyofprogramminglanguageslikePython,C++,Java,etc.Itcanprocessimagesandvideostoidentifyobjects,faces,oreventhehandwritingofahuman.Whenitisintegratedwithvariouslibraries,suchasNumpywhichisahighlyoptimizedlibraryfornumericaloperations.

READING/WRITING AN IMAGE FILE

An image is a multidimensional array; it has columns and rows of pixels, and each pixel has a value. For different kinds of image data, the pixel value may be formatted in different ways.

 #create a 3x3 square black image using a 2d Numpy array
 img= np.zeros((3,3),dtype=np.unit8)
 
 #convert this image into RGB 
 cv2.cvtColor(img,cv2.COLOR_GRAY4BGR)

IMAGE READING

The imread() function loads image from the specified file and returns it.

 cv2.imread("image')
 #reading image with grayscale mood
 cv2.imread("image",IMREAD_GRAYSCALE)
 
 #using imread for reducing size
 cv2.imread(image,cv2.IMREAD_REDUCED_COLOR_2)
 CV2.IMREAD(IMAGE,cv2.IMREAD_REDUCED_GRAYSCALE_8)

DISPLAYING CAMERA FRAMES IN A WINDOW

if event == cv.EVENT_LBUTTONUP

Image Processing

CONVERT FROM GRAY TO BINARY

  cv2.threshold(grayimage,127,255,cv2.THRESH_BINARY)

BGR TO RGB

  cv2.cvtcolor(image,cv2.COLOR_BGR2RGB

split

  B,G,R=cv2.split('image')

split HSV

 hue=[:,:,0]
 saturation=[:,:,1]
 value=[:,:,2]

resizing

  half = cv2.resize(image, (0, 0), fx = 0.5, fy = 0.5)
  bigger = cv2.resize(image, (1050, 1610))
  stretch_near = cv2.resize(image, (780, 540),interpolation==cv2.INTER_NEAREST)

LOW PASS FILTER

BLURRING

average

    cv2.blur(image,(5,5))

Gaussian

    cv2.GaussianBlur(image, (7, 7), 0)

Median Blur

   cv2.medianBlur(image, 5)

Bilateral

   cv2.bilateralFilter(image, 9, 75, 75)

HIGH PASS FILTER

smooth filter

  filterkernel1 = np.array( [[0.04, 0.04, 0.04, 0.04, 0.04],[0.04, 0.04, 0.04, 0.04, 0.04],[0.04, 0.04, 0.04, 0.04, 0.04],[0.04, 0.04, 0.04, 0.04, 0.04],[0.04, 0.04, 0.04, 0.04, 0.04]])
  cv2.filter2d(img, -1, kernel1)

sharp filter

  filterkernel2 = np.array([[-1, -1, -1],[-1, 9, -1],[-1, -1, -1]])
  ndimage.convolve(img, kernel2)

EDGE DETECTION

CUSTOM KERNEL

ndimagekernel_3x3 = np.array([[-1, -1, -1],[-1, 8, -1],[-1, -1, -1]])
ndimage.convolve(img, kernel_3x3)

DERIVATIVES

laplacian

   cv2.Laplacian(img, cv2.CV_64F)

sobel and scharr

  sobelx = cv2.Sobel(img, cv2.CV_64F, 1, 0, ksize=5)
  sobely = cv2.Sobel(img, cv2.CV_64F, 0, 1, ksize=5)
  cv2.add(sobelx, sobely)

CANNY

  cv2.Canny(img, 200, 300)

Contours

can be explained as the curve joining all the continuous points along the boundary which are having the same color or intensity .

Applications of Contours detection:

shape Analysis object detection Object recognition

1

For better accuracy , we use binary image for finding the contours so first we are going to generate the binary image

  cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)

2

apply threshold mentioned in the previous section to be able to detect corners by adding this code

   ret , thresh = cv2.threshold(grayimage ,127,255,0)

3

find out contours

  cv2.findContours()

retrieve the entire hierarchy of external and internal contours

   cv2.RETR_TREE

retrieve the most external contours

   cv2.RETR_EXTERNAL

back to the code sample, note that the findContours function returns two elements: the contours and their hierarchy. We use the contours to draw green outlines on the color version of the image

   contours , hierarchy = cv2. f indContours( thresh, cv2.RETR_TREE,cv2.CHAIN_APPROX_NONE)

   rho=1 and theta=np.pi/180.0

threshold represents line is discarded

     cv2.HoughLinesP(image,1,np.pi/180,20,minLineLength,maxLineGap)

Detecting Circles

HoughCircles

It works in a very similar fashion to HoughLines

HoughCircles has a minimum distance between a circle's centers, as well as minimum and maximum values for a circle's radius

cv2.HoughCircle(image,cv2,Hough_Gradient,1,np.pi/180,20,minRadius=0,maxRadius=0)

Draw shape on Image

line geometric

pt1 is the First point
pt2 is the Second point
Line color
Line thickness

cv2.line(img,pt1, pt2, color, thickness) cv2.line(img,(0,0), (255,255), (255,255,0), 3) cv2.arrowedline(img,(0,0), (255,255), (255,255,0), 3)

rectangle geometric

pt1 is the Vertex of the rectangle
pt2 is the Vertex of the rectangle opposite to pt1
rectangle color
rectangle thickness
Negative values

cv2.rectangle(img, pt1, pt2, color, thickness) cv2.Rectangle(img,(0,0), (255,255), (255,255,0), 3)

circle geometric

center
radius
color

thicknes

cv2.circle(img, center, radius, color, thickness)
cv2.circle(img,(0,0), (255,255), (255,255,0), 3)

ellipse geometric

center
axes is the half of the size of the ellipse main axes
angle is the ellipse rotation angle in degrees
color

thickness

cv2.ellipse(img, center,axes, angle, startAngle, endAngle, color, thickness)

text string

center
text string
fontScale
fontFace
color

thickness

cv2.putText(img, text, org, fontFace, fontScale, color, thickness)

rehamessa/Computer-vision