数显表的自动分割算法研究..doc

毕业设计(论文) 数显表的自动分割算法研究 系 别 电子信息系 专 业 电子信息工程 班级学号 5061427 姓 名 侯洁 指导教师 苗立刚 2010年 6 月15 日 数显表的自动算法研究 摘 要 数字仪表以其精度高、读取方便、易设置等优点在工业和检测领域得到了广泛的应用。但由于某些仪器仪表并没有提供与计算机进行数据通信的接口，要实现检测自动化，需要利用数字图像处理技术来实现仪器仪表读数的高速自动识别。 本文数码管显示的数字字符识别中的图像预处理进行研究。在分析字符识别的现状以及发展趋势的基础上，针对数字字符显示单元固定的特点，围绕其进行了系统的工作。 采用的是局部阈值和数学形态运算的方法，先对图像运用高斯低通滤波来处理噪声，然后对滤波后的图像进行卷积平滑，接着进行局部阈值的二值化处理，可以得出效果较好的二值化图片。而在去除图像的杂质方面，本文对图像的区域属性进行分析，运用欧拉数椭圆的偏心率来去掉杂质。 关键词：数显表分割，二值化 The Research of Automatic Identification and Segmentation on Numerical Instruments Author: Hou Jie Tutor: Miao Ligang Abstract Digital display instrument are widely used in industrial and detection system field for its high precision, convenient reading and easily setup. Data displayed by instrument need to be automatically collected. As for some instruments that have no interface to communicate with computer, in order to achieve automatic detection, image processing technique is used to collect and recognize the numerical characters displayed by instruments. So instrument recognition is very valuable. Numerical characters recognition using image processing technique and image recognition technique is studied in this dissertation. The current condition and development trend of numerical characters recognition are firstly analyzed, and then a new high speed recognition approach of numerical characters is presented according to its inherent features. Using the local threshold method and mathematical morphology operations, first of all using Gaussian low pass filter to deal with noise. Second, use the convolution smoothing to the filtered image, and followed by local threshold binarization processing. Finally, better results can be obtained binary image. Remove impurities in the image area by analysis binary image of the regional property，using the nature of the concave and convex degree, Euler number, and the e