Image_Overlaying-with-OpenCV

Adding Image over Image with OpenCv

---

The result images are stored in result directory. | _______________________________________________________________|

This project was created with pycharm environment opencv version 4.5.3 Python 3.7.11 Numpy 1.21.2

Python scripts

Initial.py

This is the main py file.

variables

bgImg_path :: Path of background images (luggage) fgImg_path :: Path of foreground images (threat)

bgImage_list :: list of Path of background images fgImage_list :: list of Path of foreground images

label :: this is concated string of background image name and foreground image name

The nested for loops call the overlay function which returns the result This result is stored in finalImg variable

cv2.imwrite will save the results in Media/processed_images directory

---

---

Adding_images.py

---

overlayImage function

This function take path of background and foreground image as input arguments And returns an image containg both foreground and background images over each other

---

variables

bgImage :: Background image (luggage image) bgImage_alpha :: b,g,r to b,g,r,a transformed backgorund image (for alpha channel)

fg_transformed :: foreground image after transformation (45 degree angle and rescaling) fgImage_alpha :: b,g,r to b,g,r,a transformed foreground image (for alpha channel) weighted_image :: Image after adding both foreground and background images

frame :: final image after again changing color from bgra to bgr

Note* we can play with alpha, beta values of addweighted to change translucency of background and foreground image

---

---

Transforming_images.py

---

img_transform function

This function takes forground image and background image path as input in its argument and returns an image after cropping its background

---

variables

img_path :: foreground image path bg_path :: background image path

lower, upper :: lower and upper limit for inRange function inRange function will keep all the white background as it is and transfrom the threat(knife) image to black color

Next with getStructuringElement a kernel is created for finding the region of interest Then a morphological operation is applied with morphologyEx I had used MORPH_ERODE (after trial and error)

to convert the image back to white pixel I deleted morph values from 255

then I performed a bitwise_and operation to get color image

Then obtained image was to enlarged as original image so a rectangular boundind box is used to crop intrested part

---

img_rotate function

This function will return an image after rotating it by 45 degree and also rescale it by half

---

variables

hh, ww :: height and width of background image img :: image from image_transform function bgImg :: background image mat :: Rotation matrix to rotate and rescale image

The reason for passing background image to this function is that addweighted function needs both overlaying images of same shape to add, So I converted the resultant foreground Image into size of background image

getRotationMatrix2D will create a matrix to rotate image warpAffine Applies a Rotation to the image after being transformed. This rotation is with respect to the image center