摘要
购物小票是消费行为的真实记录,对其进行识别有重要的现实意义。小票图像的预处理操作是对其正确分割和识别的关键步骤,预处理结果的好坏对后期的识别产生直接的影响。预处理阶段包括图像去除干扰色,灰度化,二值化,平滑去噪,倾斜校正,模糊图像还可用图像增强等处理方法。各种各样的拍摄角度和光照条件各异的小票图像,经过本文的预处理操作,可直接进行后期的字符分割和识别。
Shopping receipt is the real record of consumer behavior, and it has important practical significance to identify it. The pretreatment of the receipt image is the key step for the recognition, and the results of the pretreatment have a direct impact on the late identification. The pretreatment includes image denoising, gray scale, image filtering, binarization, tilt correction, and so on,and image enhancement and smoothing are also used in the blurred image. Through the pretreatment proposed in this paper, the variety of receipt images, with different shooting angles and illumination conditions, can be directly carried out late character segmentation and recognition.
出处
《计算机时代》
2016年第4期21-24,共4页
Computer Era
关键词
去除干扰色
灰度化
二值化
平滑去噪
倾斜校正
remove interference color
gray scale
binarization
smooth denoising
tilt correction
