使用邊界和中心之彩色商標檢索系統

Abstract

本論文研製之彩色商標檢索系統提出一個簡單又有效之彩色商標檢索方法, 首先偵測出彩色商標的邊緣. 然後計算出涵蓋所有邊緣的最小圓圓心. 以得出的圓心和邊緣為基礎可以得到四個特徵值, 包括距離統計圖, 取樣之外廓, 邊緣像素之密度以及飛鏢靶上之中靶統計分布. 基於以上四個特徵值建構出一個有效的彩色商標檢索系統.

This thesis proposes simple method for color trademark retrieval. First, the edges of each color trademark are detected. Then, the center of the smallest circle that can cover all the edge pixels is derived by a geometry method. Based on the center and edges, four features of the color trademark are extracted, these include distance histogram, sampled contour, edge pixel density and hit statistic on a dartboard. Using these features, an efficient content-based color trademark retrieval system is constructed. In order to show the effectives of the proposed system, a comparison with those methods using Zernike moment magnitudes (ZMMs) is also made, and the experimental results show that the proposed system using hit statistic on dartboard is superior to ZMMs.