隨著台灣半導體產業愈來愈進步，IC功能日趨強大，體積也愈來愈小。要確定出產的IC是否完好，IC檢測是必要的工作，從一開始的人工檢測到現在的AOI技術，可得知檢測技術方面上也是愈來愈進步，在大部分的檢測主要以接腳瑕疵、線路瑕疵做檢驗。此篇論文主要檢測的是封裝在IC上的字元，利用辨識出的字元去檢測是否有瑕疵，以及方便做分類。在演算法部分，這篇論文使用了影像處理技術定位及找出字元、投影法切割字元、類神經網路訓練法辨識字元。在定位方面，主要找出晶片的四個頂點座標，這樣可得知斜率並作旋轉。在辨識方面，以倒傳遞類神經網路演算法去學習以及辨識，其中隱藏層的神經元數需要靠錯誤嘗試法，才能得知最佳收斂的數目。在測試方面，使用了60張影像進行測試，總共1020個字，錯了47個字，字元辨識率平均結果達到95%。

The semiconductor industry in Taiwan has become more and more advanced. On the one hand, IC (integrated circuit) functions are continuously increasing, and ICs are being designed in smaller and smaller sizes. In order to determine if a manufactured IC meets regulatory standards, an inspection is necessary. From the manual inspection in the beginning to the recent AOI, it shows that the upgraded inspection technology is more and more advanced. The most important task of the inspection is to detect pins of IC and the circuit lines. In this thesis, the inspection is to detect the main packaging characters on the IC. It is to use the recognized characters to detect whether the IC is defective or not so that it is convenient to do classification. The algorithm used in this thesis includes the digital image processing (DIP) to do position and find the characters on the IC, the projection to do character segmentation and the artificial neural network (ANN) to recognize characters. In the position, it is to find four coordinates of vertex of IC, the slope is obtained and the IC can do rotation. In the recognition, it is to learn and recognize by the back-propagation neural network algorithm, in which the numbers of neuron in the hidden layer are decided by the numbers of the best convergence by trial and error. In the testing, this thesis uses 60 images for testing. The numbers of total character are 1020 and the numbers of error are 47. The average of the character recognition rate is 95%.

[1] Gonzalez Woods, “Digital Image Processing 3/e”, Pearson Education Taiwan Ltd, 2009.
[2] Chi-Nan Tsai, Face Detection in Color Image Using Wavelet Neural Network, Chaoyang University of Technology, Dept. of Computer Science and Information Engineering, Taiwan, 2004.
[3] Tai-Chuan Huang, Implementation of Three-Dimensional Space Infrared Trajectory Capturing and Its Application to Interactive Whiteboard, National Cheng Kung University, Dept. of Engineering Science, Taiwan, 2011.
[4] Chun-Che Fung, “A Review of Evaluation of Optimal Binarization Technique for Character Segmentation in Historical Manuscripts”, Proc. of the 3th International Conference on Knowledge Discovery and Data Mining, pp.236-240, Jan. 2010.
[5] Wei-Ying Chen, Design and Implementation of Real-Time License Plate Recognition with Low Cost Based on Embedded System, National Cheng Kung University, Dept. of Engineering Science, Taiwan, 2011.
[6] Dongju Liu, “Otsu Method and K-means”, Proc. of the 9th International Conference on Hybrid Intelligent Systems, pp.344-349, Aug. 2009.
[7] Tsung-Yen Chen, Design and Implementation of The Face Tracking and Recognition System, National Cheng Kung University, Dept. of Engineering Science, Taiwan, 2007.
[8] Bo-Jhih Hu, Implementation of Neural Network and Its Application to Handwriting Recognition System Using Touch Panel, National Cheng Kung University, Dept. of Engineering Science, Taiwan, 2011.
[9] Chun-Lin Huang, Design of Real-Time Object Tracking System Using CamShift Algorithm, National Cheng Kung University, Dept. of Engineering Science, Taiwan, 2009.
[10] Ti-chia Hsu, Euler Angles Representation of N-dimemsional Homogeneous Transformation, National Cheng Kung University, Dept. of Manufacturing Engineering, Taiwan, 2007.
[11] Zhang X., Liu X., and Jiang H., “A hybrid approach to license plate segmentation under complex conditions”, The 3rd International Conference on Natural Computation, pp. 68-73, August 2007.
[12] Zhang Y., and Zhang C., “A new algorithm for character segmentation of license plate”, 2003. Proceedings. IEEE Intelligent Vehicles Symposium, pp. 106-109, June 2003.
[13] Meng-Chiou Liao, High Performance Fingerprint Identification and Its Implementation, National Cheng Kung University, Dept. of Engineering Science, Taiwan, 2009.
[14] 廖詩芳，符合人眼色差知覺之均勻色彩空間，國立中央大學光電科學所碩士論文， 2007年。
[15] 王彥棋，以邊為基礎之衛星影像中的車輛偵測，國立中央大學資訊工程所碩士論文， 2004年。
[16] 黃偉銓，應用影像處理技術於統一發票之號碼自動辨識，國立台灣科技大學高分子工程系碩士論文，2008年。
[17] 葉本源，適用於台灣各種車輛之車牌辨識系統，中原大學電子工程所碩士論文，2006年.
[18] 王進德，類神經網路與模糊控制理論入門與應用，全華科技圖書股份有限公司，台灣，2007年。