拉普拉斯差分算子同步降噪空间滤波器设计

Design of synchronization dimensional structure filter for Laplace operator

为了在继承拉普拉斯算子高实时性的前提下提高图像分割质量,设计了一个可与拉普拉斯算子差分计算同步进行滤波工作的空间滤波器。以被检测边缘的连通域为单位,建立连通域点集及收录二值化数据的树集,引入面积阈值进行滤波。运用游程编码的思想作为树集设计逐行扫描的数据录入方法,使数据录入与拉普拉斯差分运算同步进行,以保证高实时性。边缘检测时通过降低差分计算的颜色阈值获取更为完整的检测结果,同时利用空间滤波器去除由颜色阈值的降低所引发的大量小面积噪声。试验结果表明:引入空间滤波器的拉普拉斯算子可在保证高实时性的前提下获取低噪声的且更加完整的边缘检测结果。同步空间滤波器的引入,可使原算法在低颜色阈值条件下获得高质量的图像分割结果,且该滤波器的行扫描数据录入方式可以保证高实时性。

In order to improve the quality of Laplace operator image segmentation at a high real time capability this paper presented a synchronization dimensional structure filter for Laplace operator. After constituting interconnected domain volume, a square threshold is introduced. Using the Run-Length coding method, implementing the synchronization of volume date-recording and Laplace difference, the filter keeps a high real time capability, The synchronization filter filtrates the thin square noise when the color threshold is depressed to obtain a more intact detection. The result indicates that the dimensional structure filter edge method can obtain a more intact and less noise image segmentation. The introduction of the synchronization dimensional structure filter makes the Laplace operator improve the segmentation quality with a high real time capability by the original arithmetic.