形状の類似性評価方法

概要

論文の詳細を見る
At first, we recognize 2 dimensional object shape with the convex hull and treat feature to be provided by this. And we use feature provided from picture element of object, too. This feature is provided orthograph of picture element to X/Y axis. It is feature extracted from waveform which is provided after coordinate transformation around center of gravity of convex hull. We regard feature by these as point of N demensional space. And we compare similarity of object by this distance with a human sense. Because a human sense is different with individual, we perform a questionnaire. Because a human sense is defferent with individual, we perform a questionaire. We make fuzzy membership function using effect got by a questionnaire next. And we input feature distance and get similarity. Finally we test picking experiment with CDD camera, and confirm validity of estimation approach.
山梨大学の論文