一种新的快速字符定位算法研究

准确定位字符区域是字符识别的第一步,针对高速运动拍摄图像对比度差、噪声干扰严重、字符区域范围变化大、要求在线识别且时间短等特点,研究了一套特殊的数学形态学和图像滤波算法,算法利用字符局部信息对区域内的字符进行二值化处理,后进行连通运算和数学形态学闭运算,使相邻字符连成一个字符串区域,最后根据侯选区域的先验信息,快速定位出被识别字符的准确位置.研究表明,在高速运动中拍摄80*60mm大小的物体,图像大小为640*480,利用ROI区域,可以实现100%的目标区域精确定位,定位时间小于20ms,为后续字符的准确分割和识别奠定了良好的基础.

Research on a New High-speed Character Locating Arithmetic

Locating the character exactly is the primary step of character recognition.To some features of photos at high-speed,such as low in contrast,noise interference,high range of characters variety,and requirement of online recognition in a short time,a special algorithm on math morphology and graphic filtering has been proposed,it performs bivalue of the character in the area according to its local features,and then performs the connect operation and close operation of math morphology to connect the adjacent characters into one whole character,and finally locates the characters in high speed according to its Priori Information.Research shows that when you take a photo of an object(80*60mm) in high speed,the size of the picture will be 640*480.By dealing with the ROI area,you can get an accurate 100% location of the target area,and the time of locating is less than 20ms.This new algorithm lays good foundations for the subsequent operation,the division and recognition of the character.