面向彩印标签的手机数码识别及应用

针对Android手机拍摄标签数字码实时识别的需要，以及拍摄图像具有分辨率低、亮度不均匀、背景复杂等特点，由手机拍摄直接获得灰度图像后，通过数字区域投影定位，获取只包含数字的图像;采用二值化进行灰度图像到黑白图像的变换;通过投影及归一化处理进行数字码图像分割，并对每个数字码图像进行细化获取细化数字码;基于统计学抽取数字码的特征;建立数字码模式特征后，采用最近邻域判别函数进行数字码识别，取得了良好的识别效果。

Numeral recognition and application of mobile for color-printed label

For the need of real-time identification of numerical label code and the problems of low resolution, uneven brightness and complicated background of the grayscale images taken directly by Android mobile, the numerical images are obtained after the grayscale images are projected and then the numerical area is located. The grayscale images are transformed into the black-and-white images. The numerical images are split by using projection and normalization processing. And each numerical image is thinned to obtain the thinned numerical code. Then the features of each numerical code are extracted based on the statistical method. After each numerical code pattern is established, the numerical code is recognized by using the nearest-neighbor classification function which achieves very good results.