结合Prewitt算子和旋转投影的纸币图像倾斜校正法

纸币图像的倾斜会严重影响其中字符识别的精度,为此需要在字符识别前对纸币图像进行倾斜校正.基于Prewitt算子和旋转投影提出一种新的纸币图像倾斜校正法.该方法采用水平边缘检测、垂直边缘检测以及旋转投影求取纸币图像的水平倾斜角度和垂直倾斜角度,然后对水平倾斜的纸币图像进行双线性插值旋转校正,对垂直倾斜的纸币图像进行双线性插值错位偏移校正.另外,考虑到实际应用中获取的纸币图像的倾斜情况,为保证方法的实时性,水平倾斜角和垂直倾斜角均限定在10°以内.校正结果表明此方法效果较为有效.

Correction method of banknote image slant by using Prewitt operator combined with rotation projection

The slant of banknote image may have a serious influence on the precision of character recognition.Therefore,banknote image slant correction is necessary before the character recognition.Based on Prewitt operator and rotation projection,a new correction method of banknote image slant was proposed.In this method the horizontal edge detection,vertical edge detection,and rotation projection were adopted to obtain the horizontal and vertical slant angles of banknote image.Then the bilinear interpolation rotation correction was performed for horizontally slanted banknote image,and the bilinear interpolation offset correction was made for vertically slanted banknote image.In addition,considering the slant condition of banknote image in practical application of this method,the both in order to ensure the real-time property of the proposed method,horizontal angle and vertical angle were confined inside 10 degrees.Correction result showed that the proposed method exhibited a better effect.