基于Tesseract开源OCR引擎的证件识别系统的设计与实现-计算机技术专业论文.docxVIP

验中的证件识别器。 本文针对特定格式的类身份证性质的证件，通过对图像源的预处 理，给予特定样本的训练，实现一种基于 Tesseract-ocr 的包含不同字 体汉字数字的证件格式化 OCR 识别系统。Tesseract 是一个开源的 OCR 引擎，支持汉字等等绝大多数文字的识别，并且提供了一系列 训练的方式，可以自行训练出需要的训练集。针对不同的场合进行优 化。也支持大量的参数调优，例如行距，字距等等信息。 在本文的实验中通过一些技术，使得 TesseractOCR 的识别效率 以及识别的精确度大幅提升，分析的过程大致是通过证件的固有格式 进行区域的划分而在对不同区域的识别上采用不同的参数以及不同 的训练集。例如对性别的识别，所采用的训练集在训练时只训练了两 个字男和女，这样的训练结果虽然带有一些局限性，但是应用于大多 数的场合都是能提升一定的效率和准确性。 本文的创新之处就是在于摆脱了价格高昂的商用 OCR 软件，而 采用了开源的 TesseractOCR 可以降低整套系统的成本并且在效果上 满足需求。 关键字： OCR 软件，数字图像处理，证件识别 ，tesseract The Design And Development Of Identification Recognizing System Based On Tesseract Abstract Identity card recognizing is nowadays more common. For example when you check in or occur to some traffic accident ,identify card should be checked to show identity . When these process being done ,lots of work always done by human hands .And it cost too much time . These check process can be used for preventing defective card come into social application . With the improving of network tech , image processing tech ,ocr tech more and more recognizing process can be done by computer .Because of the high efficiency of computer ,work can be quickly finished . Getting source picture is another important thing .Best choice of device is scanner. It can produce clear and high definition pictures .But in most situation camera is the only available .Using camera , a lot of problem come to actual practice .And in this paper we present some method to improve the quality of source picture . In correcting twisted pictures ,these paper present the theory and the tool which help in fast developing and efficient processing. The most important thing in these paper is using and open-source ocr engine .That means the cost is very low .And it can be somehow useful. In this paper I customized some special traineddatas for recognizing different source pictures.These new traineddatas are designed having limited word list and frequent word list and adept at recognizing at ethnic ,gender or address. One more new