【作者】 何金成；

【导师】 张性雄；

【作者基本信息】 福建农林大学 ， 农业机械化工程， 2003， 硕士

【摘要】 在农机测试中，作为测量数据存储形式之一的纸面记录曲线，由于采用手工数据处理方法，处理效率低、精度不高、处理工作强度大。为改变此现状，本文提出利用扫描仪把纸面记录曲线转化为数字图像，然后尝试运用图像识别的方法对纸面记录曲线进行自动化处理，把数字图像中的曲线数据及标定信息识别出来，再对提取的数据进行分析计算并得出结果。该方法的关键问题在于把记录曲线从纸面背景中正确分离、识别，并对纸面有污点、污线的记录曲线有较高的正确识别率。本文提出了基于三基色的类判别阈值分割法来获取图像中目标与背景的分割阈值，并提出基于纸面记录曲线图像中的记录曲线的特征的双边缘跟踪细化算法和曲线中轴搜索细化算法，来识别分离记录曲线，较好的解决了本方法的关键问题。更多还原

【Abstract】 In this paper, a new method to digitalize testing curves recorded on paper, is proposed to resolve the existing problems with processing the manual paper-based curves. A scanner is used to gain the image of paper-based testing curves. A special computer program-an image-processed program is developed, which used to recognizeand segment the curves and the lines of calibration in its background. A algorithm of Ostu method with maximum variance between two classes based on three primary-color is introduced to gel the threshold of curves image segmentation. And a fast thinning algorithm based on double edge detection and the character of the curves is also introduced, which could extract the curves reliably and effectively as the same time as image thinning. This method is proved efficient by practical application.更多还原