how to detect defect in pcb using opencv python

Question

I am working on project in which i supposed to detect defect in PCB i have already tried image subtraction method. It detects the defect missing components in PCB but it is not giving correct output even when PCB is not defect. it means it show that there is missing component in PCB even the component is present

here is me code

g_o_img = cv2.cvtColor(o_img, cv2.COLOR_BGR2GRAY)
g_def_img = cv2.cvtColor(def_img, cv2.COLOR_BGR2GRAY)
score, diff = structural_similarity(g_o_img, g_def_img, full=True)
diff1 = (diff*255).astype("uint8")
cv2.imshow("diff", diff1)

ssim = format(score)
print("SSIM : "+ssim)
l_b = np.array([0])
u_b = np.array([50])
mask = cv2.inRange(diff1, l_b, u_b)
kernel = np.ones((3, 3), np.uint8)

erode = cv2.dilate(mask, kernel)

cv2.imshow("mask", mask)
cv2.imshow("erode", erode)

Answer 1

You have to do some preprocessing before background subtraction. You have to detect the PCB only (without the background) and make sure the perspective is always the same. If the perspective is not the same do some Perspective Transformation to crop the PCB out of the image (Check here). Then convert the image to a Color Space that is more robust to light and shadows (Check here). After that you can do background subtraction followed by Thresholding (Check here )to get a binary image. Finally you can do Morphological Transformations to remove the false positives (Check here) and leave only the differences between the images .