Week 8

Testing Threshold Template Matching

Introduction

This week’s blog post will cover the testing done on the differences found in Week 6. This should only get the patches that contains differences.

Code

The first function getBestMatch() finds the best matching patch from the second image. It returns the coordinates of the patch the match value (between 0 and 1). The second function getBestPatches() searches through the list of patches found in Week 6 and returns all the patches that have a match value (from getBestMatch()) lower than a theshold [1].

def getBestMatch(img, patch):
    patchSize = patch.shape

    gImg = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    gPatch = cv2.cvtColor(patch, cv2.COLOR_BGR2GRAY)

    # cv2.TM_CCOEFF_NORMED or cv2.TM_CCORR_NORMED
    result = cv2.matchTemplate(image = gImg, templ = gPatch, method = cv2.TM_CCOEFF_NORMED)

    (_, value, _, (x, y)) = cv2.minMaxLoc(result)

    return ((x, y), value)

def getBestPatches(sourceImg, checkImg, patches, threshold = 0.5):
    bestPatches = []
    for (x, y, w, h) in patches:
        patch = sourceImg[y : y + h, x : x + w]
        ((mX, mY), matchValue) = getBestMatch(checkImg, patch)
        if matchValue < threshold:
            bestPatches.append((x, y, w, h))

    return bestPatches

cv2.TM_CCOEFF_NORMED

These are the differences found between pcb1.jpg and pcb2.jpg using cv2.TM_CCOEFF_NORMED with a threshold of 0.35.

cv2.TM_CCOEFF_NORMED - 0.35

cv2.TM_CCORR_NORMED

These are the differences found between pcb1.jpg and pcb2.jpg using cv2.TM_CCORR_NORMED with a threshold of 0.8

cv2.TM_CCORR_NORMED - 0.8

Conclusion

In the results above, cv2.TM_CCORR_NORMED provided the best results. The other method detected more patches as being different. To solve this issue, the images should be normalised to look as similar possible. This could be achieved by finding the edges of the images and only using those to find different features. This will be futher explored in a future blog post.

References

[1] OpenCV 3.3.1 Tutorials, 2017, [Online]. Available: https://docs.opencv.org/trunk/d4/dc6/tutorial_py_template_matching.html. [Accessed: 2017-11-04]