智慧型字卡影像辨識系統設計

Abstract

本篇論文的研究目的是希望能製做出一系統，能使童伴機器人與孩童進行互動，並從中學習單字。系統被設計成三個步驟，包含抽取物件的區域、抽取字元以及字元辨識。本篇論文有主要兩個貢獻，第一部分是特定顏色移動物的辨識。第二部分是在不同大小、傾斜以旋轉中的字元辨識。在特定顏色移動物的辨識中，一般所知的是顏色辨識與移動物辨識兩步驟。本篇論文在此使用類神經演算法將一般所知的兩個步驟改成同時進行。此外在針對文字辨識中，我們使用熟知的標楷體來做辨識，此系統可以在不同位置、大小、傾斜程度以及任意的旋轉角度中，達到成功的辨識。最後，三個實驗將會驗證系統的成功開發。

The purpose of this thesis is to build up a system for children to learn words in an interactive way. The system is designed in three steps, including potential object localization, character extraction and character recognition. The main contribution of this thesis includes two parts, which are the moving object detection and the image recognition. The moving object is an word card in special color and the image to be recognized is a character. To detect the moving word card from a sequence of images, an artificial neural network is proposed to extract the color and detect the word card simultaneously. After the word card is detected, a scheme based on a set of concentric circles is applied to extract the features of the character on the word card. From the features, another artificial neural network is designed for character recognition invariant to the translation, rotation and scaling of the word card. Finally, the success of the developed system is verified by three experiments.