借助形态学实现图像过渡区提取与分割

红外图像边缘模糊、噪声点多,而传统的过渡区提取算法需计算梯度因而对非规则细节和噪声敏感,尤其当对比度较低时极易导致提取的过渡区发生偏离,影响分割阈值的判定和分割结果的准确性。因此提出了一种基于形态学的过渡区提取与分割算法。结合形态学理论,首先设计级联滤波器进行平滑处理,再利用tophat变换调节目标与背景的对比度以突出目标,最后采用EAG方法进行过渡区提取与分割。Matlab仿真实验表明,该方法能够分割出目标主体,且目标形状保持良好。

Extraction and segmentation of image transition region based on morphology

The infrared image has blur edges and many noise points.Traditional methods need to calculate the grads,so they are sensitive to unregular details and noises.When the contrast is low,departure may appear for extracted transition region,thus will affect the judgement of segmentation threshold value and the segmentation result.Therefore,we put forward a new method for transition region extraction and segmentation based on the theory of mathematical morphology.A cascade filter is designed for smoothing processing;and then tophat is used to adjust the contrast to stand out the object.Finally, EAG is used for the extraction of transition region.Matlab simulation experiments showed that the proposed method can segment object's main body,and keep the figure well.