Zhang并行细化算法的改进

Improving of Zhang parallel thinning algorithm

手写数字字符,由于书写时的无约束性,字体太小或书写时笔墨的扩散,导致出现较大的黑块,当进行细化时该黑块就会被细化为线或者点,这在字符识别中是非常不合理的,将严重影响系统的识别性能。为了减少细化时所带来的信息丢失,采用了设定阈值的判别方法,提出了基于行像素和最大值与列像素和最大值的交点处及其邻域内像素取反的方法对图像进行处理,在细化时可以有效地保持字符的形状,一定程度上提高了细化效果。

To handwritten digital character, when the font is too small or the ink proliferation, resulting in larger black lump in thedigital image. This lump can be a line or a spot by the thinning, this is very unreasonable in character recognition, and will seriously affect the recognition result. In order to reduce the loss of information by the thinning, the assessment method of setting threshold is used. Proposing the plan that flips the value of intersection point that the row pixels, sum is maximum and the column pixels, sum is maximum. It can effectively keep the shapes of the characters, the result effect is improved.