objects detector infrastructure based on classic method (sharpening + laplacian + thresholding)

The program load an image and manipulate the output. This program not classify the object, just markers its edges.

libraries

• cv2
• numpy
• skimage.exposure

example

There is a car image from google search just for the test, you can change the image and make sure to change its path in the program.

parameters

• marker_width is the width of the points markering the edges
• threshold is the intensity of the pixel after applying convolution with the laplacian kernel

FYI

You can change the sharpening kernel and the laplacian kernel to whatever you want. Suggestion:

• sobel
• canny (output the edges)