一种改进的二值图像像素标记算法  

An Improved Pixel Labeled Algorithm in Binary Image

作　　者：王亮[1] 黄飞[1] 冯少彤[1] 聂守平[1] 王鸣[1] 

机构地区：[1]南京师范大学光电技术江苏省重点实验室,江苏南京210097

出　　处：《南京师范大学学报（工程技术版）》2006年第2期23-25,29,共4页Journal of Nanjing Normal University(Engineering and Technology Edition)

摘　　要：像素标记算法是图像处理中提取目标特征量的一种有效方法．分析了一般像素标记算法的不足,提出一种改进的像素标记算法．该算法无需对图像进行预处理,针对二值图像中像素分布的特点,仅需对图像进行一次扫描,通过特殊的目标边界像素标记,就能收集足够的信息,经过一定的分析处理即可得到图像中目标的准确计数和边界信息．由于准确的目标计数是目标特征量提取的关键,在此基础上即可方便快捷地获得目标的面积、周长和质心坐标等参数,便于图像的后续处理．计算机模拟结果表明该算法具有很强的通用性,且因为仅需一遍全图扫描使其分析速度大大提高,可以广泛应用于图像分割领域．The pixel labeled algorithm is an effective method to obtain the objects＇ s feature in image processing. Some limitations of the normal pixel labeled algorithm are analyzed and an improved one is presented in this paper. The new algorithm reduces the number of image scanning to only once and it can collect enough information during the scanning. After some necessary processing, the correct number of objects and the edge information can be obtained. Consequently, the object＇ s area, perimeter and coordinate value of center of mass are also conveniently obtained, which is essential to some further image processing. Computer simulation results show that this algorithm is practical and much faster; therefore, it can be widely used in image segmentation field.

关 键 词：图像处理 像素标记 目标计数 边界信息 

分 类 号：TN911.73[电子电信—通信与信息系统]