基于测地线活动轮廓模型的图像除噪和增强

Denoising and Enhancement Image Based on Geodesic Active Contour Model

讨论曲线演化的活动模型去除图像噪声的性质。利用不同尺度小波变换系数的边缘映射指示图像的边缘。把不同尺度下的边缘映射经复合设置成曲线演化方程的边缘指示函数。为了防止边缘尖点被平滑,采用一种新的数值离散迭代格式求解曲线演化方程。这样使曲线演化方程在均匀区域能更好地平滑噪声,边缘点得到保护且不被模糊。通过对图像的仿真试验说明提出的方法在图像除噪方面有良好的效果。

The property of the active contour model for image denoising was studied. At each resolution level, the image edges are estimated by gradient magnitudes obtained from the wavelet transform coeffi- cients. New edge indictor function is constructed by combining gradient magnitudes obtained from the wavelet transform coefficients at different resolution levels. New discrete numerical scheme is used. The edge, salience and local detail can be preserved well in images. The image denoising numerical results demonstrate the good performance based on modifying geometric model.