For recognition, the front or trailing edge of projection of binary image on horizontal axis basically shows linear shape, so least square theory is applied in tilt correction.
识别过程中，数码管二值化后的图像在水平轴上的投影，其上升沿或下降沿基本呈现直线的形状，尝试把最小二乘法应用于倾斜校正。
For tilted characters only in horizontal direction, even without tilt correction, this method can achieve better segmentation.
这种分割方法对只在水平方向倾斜的字符，即使不经过倾斜矫正，也能达到较好的字符分割效果。
应用推荐