基于水平集和视觉显著性的植物病害叶片图像分割

Plant Disease Image Segmentation based on Level Set and Visual Saliency

摘要:

为了提高植物病害叶片图像分割的准确性和效率,本文提出了一种基于水平集和视觉显著性的彩色图像分割方法。首先采用基于小波变换的显著性检测算法得到活动轮廓模型中曲线演化的初始位置,并构造一个基于显著区域的图像活动轮廓模型,再设计一个向量值图像的边界检测算子,引入到距离正则化水平集演化的改造中,以构造一个初始化轮廓更灵活,演化速度更快,目标分割更精确的新的水平集能量泛函。最后的实验对比表明,本方法具有较好的叶片病害部位分割效果。

Abstract:

In order to improve the accuracy and efficiency of plant disease image segmentation, this paper propose a color image segmentation method based on level set and visual saliency. Firstly adopt a saliency detection algorithm based on wavelet transform to get the initial position of curve evolution in the active contour model, and construct an active contour model based on salient regions. Then design an edge detection operator of vector-valued image, and introduce it into the reconstruction of distance regularized level set evolution, to construct a new level set energy functional with a more flexible initial contour, faster evolution speed and more accurate object segmentation. Finally experimental comparisons demonstrate the proposed model has a good leaf disease segmentation effect.