Design and Implementation of an ASM-Chart-based Document Image Analysis System

• Main Authors: Jian-syuan Hong, 洪健軒
• Other Authors: Chen-Mie Wu
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/85686151417135901559

碩士 === 國立臺灣科技大學 === 電子工程系 === 101 === This thesis is related to the design and implementation of an ASM-chart-based document image analysis system. The research work includes five parts: The first part is about software design of the binarization algorithm for digital images. The second part is related to the analysis of the document page layout. By using connected component labeling and smearing algorithms, the textlines are segmented and classified into the ASM chart and text regions. The third part is to extract the text in the ASM chart first. Then the arrowheads are also extracted by using the erosion and dilation methods. After linking the broken lines the ASM chart is processed by morphological reconstruction and thinning algorithms to extract the various symbols. Finally the line segments in the ASM chart are extracted to build the connective relations among these symbols. The fourth part is to merge the textlines into paragraphs and create a logical structure of the document which consists of both the text regions and ASM chart. The fifth part is to design an utility program for testing and validating the output data structure of the ASM-chart-based document image analysis system. And, finally, the run-time performance of the sub-programs are also shown. On the whole, this thesis has accomplished the software design of an ASM-chart-based document image analysis system. Through using the object extraction technology the ASM chart information is analyzed to build the logical structure of the document image. Besides, the related technologies can also be applied to the analysis of the flow-chart-based documents.