边缘保留自适应提升形态小波及其在甲藻特征提取中的应用

摘要

Adaptive lifting morphological wavelet has the ability of multiscale decomposition of image based on its local characteristic and geometric structure. In this paper, an edge reservation adaptive lifting morphological wavelet is put forward in order to deal with the tough task for extraction of faint sulcus of dinoflagellates with blurred edge and low contrast. Adaptivity is successfully introduced into the update lifting by leaving untouched or adopting sharpening filtering at edge and smoohing filtering at other region of image, which preserves the edge at different resolution image decomposed by this morphological wavelet. Experiments demonstrate that contrast of sulcus of dinoflagellates to the whole image are improved in muhiscale decomposition image by this morphological wavelet. Thus it is more favorable for the extraction of sulcus compared to its original grey-level image.%自适应提升形态小波将多分辨率分解和图像空域中目标的几何结构和局部特征相结合,可以满足特定图像处理任务的需要.本文针对甲藻类细胞显微图像边缘模糊、对比度低等不利于甲沟特征提取的困难,构造1种边缘保留自适应提升形态小波.该小波将图像的边缘作为重要的局部特征,更新提升根据是否是边缘像素选择保留、锐化滤波或平滑滤波,使得分解后的近似图像中,图像边缘得以保留,而灰度变化缓慢的区域得到了平滑.实验结果证明:甲藻细胞经边缘保留自适应提升形态小波分解后的近似图像,在一定程度上突出了目标边界,较之原始图像,更有利于甲沟特征的提取.