Template Matching Based De-Clumping Of Planktons In Clumped Microscopic Plankton Images
Distribution of planktons in ocean provides an important clue on marine ecosystem health. Diverse automatic analysis methods have been proposed to recognize plankton species, but only a few of them dealt with de-clumping of clumped planktoncells even though separation into individual cells is necessary before recognition. Most of de-clumping methods assumes that individual cell has a certain shape (usually circular or round) and uses circle fitting or concavity detection based separation approaches. However many plankton species neither have circular-like shape nor bear concavities in clumped boundaries. In this paper, we propose a 2-D weak affine invariant template matching method tailored to de-clumping of harmful plankton species found in Korean peninsula.We tested the proposed method with many samples and obtained almost perfect results.