多维视觉图像敏感区域智能标记方法仿真

Simulation of Sensitive Area Intelligent Marking Method for Multidimensional Visual Image

多维视觉图像敏感区域的标记在图像分析中是十分重要的步骤,高效的敏感区域标记方法能够有效提高多维视觉图像处理速度。采用当前方法进行多维视觉图像敏感区域标记时,因受敏感区域图像复杂度的影响,当图像敏感区域面积较大时,需要对各个目标像素点进行邻域判断,标记效率极大下降。为解决上述问题,提出一种基于分水岭的多维视觉图像敏感区域智能标记方法。上述方法首先通过计算获得多维视觉图像不受反射光干扰的分量梯度,将其融合于不同尺度结构元下开闭重建的多维视觉梯度图像,对多维视觉梯度图利用最大类间方差算法自动获取阈值,用得到的二值标记图像对原始多维视觉梯度图进行强制标记,确定多维视觉图像敏感区域的边缘信息,引入水平集方法通过符号压力函数来指引多维视觉图像敏感区域结构曲线演化方向,实现对多维视觉图像敏感区域完全分割,完成敏感区域智能标记。实验结果表明,所提方法多维视觉图像分割精度较高,有效避免了等效标记的出现,具有很好的鲁棒性。

An intelligent marking method for sensitive region of multidimensional visual image based on watershed is proposed. Firstly, the component gradient of multidimensional vision image without interference from reflex light is obtained by calculating, then it is fused into multidimensional visual gradient image of open and closed reconstruction under different scale structure elements, and the threshold for gradient map of multi-dimensional vision is automati- cally obtained by using Otsu algorithm. And then the gradient map of original muhidimensional vision is compulsively marked by two-value marker image obtained to determine edge information of sensitive region of multidimensional vi- sion image, meanwhile the level set method is introduced and the evolution direction of structural curve of sensitive region of multidimensional vision image with sign pressure function is guided. Finally, the complete segmentation of sensitive region for multidimensional vision image is achieved and intelligent mark for sensitive region is finished. The simulation results show that the mentioned method has high segmentation accuracy for multidimensional vision image. It can effectively avoid the appearance of equivalent marker and have great robustness.