Scale Invariant Template Matching

DONGWOOK LEE·2021년 7월 13일
0
post-thumbnail

Scale Invariant Template Matching is a Template Matching method which enables the computer to detect the template object in the test image even if the size of the object in the image differs with the one in the template. Shown above in the thumbnail image is one of the examples where the pure Template Matching method cannot detect the template object in the test image.

However, in Scale Invariant Template Matching, we allow the computer to figure out the existence of the template object in the image using greedy way. The reason I say 'Greedy Way' is because the computer simply changes the size of test image in loop. Let me show you the demonstration below.

Given the template image left, the computer tries its Template Matching with differently sized test images. As the gif image above implies, Scale Invariant Template Matching takes even much computation time due to its double loop - which consists of Sliding Window and the test image size adjustment.

Nevertheless, this method overcomes the Scale Variant property of normal Template Matching method. Using this method, you can find the object in broader range!

profile
AI/ML STUDENT

0개의 댓글