Abstract: In order to restore the contaminated PDF417 barcode image, noise should be filtered before recognizing the barcode image. A combination of morphological filtering method based on 2-D entropy is proposed. Open-closing filter based on omnidirectional multi-scale structural element is used to filter noise of the original image respectively. Weights are determined by 2-D entropy changes of each individual filter before and after filtering. Outputs of each individual filter are combined. Experiment shows that, method in this paper can effectively filter noise and preserve the local features and details of image. Recognition rate of PDF417 barcode image is improved.

Key words: PDF417 barcode, morphology filtering, 2-D entropy

摘要： 为了更好地还原被污染的PDF417条码图像，在条码图像识别前应对其进行去噪。提出一种基于二维熵的组合形态学滤波方法，采用基于全方位多尺寸结构元的开-闭滤波器分别对原始图像进行滤波，以各单个滤波器前后图像的二维熵变化大小作为权重，将各单个滤波器的输出进行组合滤波。实验表明，采用该方法可以有效去除噪声并保留图像的局部特征和细节，提高PDF417条码图像的识别率。

关键词: PDF417条码, 形态学滤波, 二维熵