2.6.3. Shape analysis and grading
Fig. 8 illustrates the entire process of oyster shape analysis and grading algorithm using the proposed TAC similarity measure approach. In order to perform accurate shape analysis, a turn-angle distribution model (called type model) for each shape category is created and stored in the database as a reference for measuring similarity. First, an oyster with a major axis close to horizontal (Fig. 6) is selected and its turn angle function is used as a standard model. Then, using the proposed TAC similarity measure, this standard model is used to ‘‘align” and shift the turn angle function of each test oyster contour to eliminate its turn angle shift caused by rotation. The resulting aligned turn angle function is called the uniform model. The uniform model of each test oyster is then compared against all three type models (one for each shape category in this work), again using the proposed TAC similarity measure, to determine the best matched shape category (the smallest TAC similarity measure of the three) to determine the test oyster’s shape category.