为准确地定位出目的地址块 ,针对中文手写信封具有的明显特点 ,提出了一种用于中文手写信封的目的地址块定位方法。二值图像在经过噪声滤波等预处理后 ,提取出连通域。用模式分类的方法将非文本特性的连通域去除 ,并由文本块合并成文本行。在已提取的文本行基础上定位出目的地址块。对 5 0 0个采集于邮政分拣机的实际样本的提取率和平均 CPU时间分别约为 85 %和 3.7s。结果表明该方法是有效的并且花费较少的 CPU时间 In order to accurately locate the destination address block, aiming at the obvious characteristics of Chinese handwritten envelopes, a method for locating the destination address block for Chinese handwritten envelopes is proposed. After the binary image is pretreated by noise filtering, the connected domain is extracted. Use the pattern classification method to remove the non-text features connected to the domain, and the text block into a text line. Based on the extracted text lines locate the destination address block. The actual sample extraction rate and average CPU time for the 500 collected in the post sorter were about 85% and 3.7 s, respectively. The results show that the method is efficient and consumes less CPU time