OCR中一种基于最小一乘法的连通域去噪方法

Method of connected domain denoise based on least absolute deviation in OCR

提出了一种在OCR过程中利用最小一乘法分析连通域大小从而实现去噪功能的方法.在图像中存在少量颜色与提取信息相似且大小具有明显差异的噪声的前提下,采用最小一乘法取代传统的最小二乘法对所有连通域的大小进行分析,从而获得需要的连通域阈值,实现去噪功能.数据结果显示,在规定前提下该方法比经典的最小二乘法所确定的范围更为准确;实验结果显示,该方法能够达到去噪效果.

A denoise approach using analyzing size of connected domain by least absolute deviation is proposed in process of OCR. There is a little color which is similar to the needed information but different from it in size. Instead of the least square method, the least absolute deviation is proposed to analyze the size of the connected domain,get the threshold of the connected domain and denoise at last. The data results demonstrate that this method compared with the classic least squares method is more accurate under the precondition. The experimental results demonstrate that this method can achieve the desired effect.