摘要
针对高速运动图像的目标识别问题,首先运用数学形态学方法进行预处理,然后采用阈值分割对待识别图像进行二值化处理,最后,对分割后的图像进行链码跟踪,利用Hu不变矩识别运动目标。试验结果表明,通过设计特殊的结构元素,利用数学形态学的Top-Hat算子,可以对高速运动图像进行有效分隔,利用Hu不变矩能够快速准确的识别空间运动目标。
Aiming at object recognition of fast moving image, firstly, the mathematic morphology preprocess is applied to the image and the image is segmented by using threshold to change it into black and white image. Secondly, the chain code of the segmented image is tracked and the moving target can be recognized using Hu moment invariant. The results of experiment show that a fast moving image can be segmented effectively by using top-hat operators when its structure dement is designed specially, and a space moving object can be recognized quickly and exactly by using Hu moment invariant after an effective segmentation method is applied.
出处
《工程图学学报》
CSCD
北大核心
2006年第6期74-77,共4页
Journal of Engineering Graphics
基金
广东省自然科学基金资助项目(B6305568)
广州市天河区科技攻关资助项目(051G029)
广州市科技攻关资助项目(2006Z3-D3151)
关键词
计算机应用
图像分割
数学形态学
目标识别
computer application
image segmentation
mathematic morphology
object tracking
