利用对称动态轮廓模型自动检测图像最优镶嵌线

Detecting the Optimal Seam Line Automatically in Image Mosaic with Twin Snakes Model

提出利用对称动态轮廓模型方法自动检测多景影像的最优镶嵌线。该方法将影像间色彩相似性和纹理相似性、公式化为动态轮廓线的能量函数。利用这种方法能够自动找到一条影像表面色调和纹理差异更小的镶嵌线,在轮廓线的选取和迭代过程中,结合了影像亮度和纹理差异的信息,检测出的镶嵌线优于仅用影像的梯度信息检测出的镶嵌线。研究以两景相邻影像自动寻找镶嵌线为例进行了实验,取得了良好的效果。

A new approach based on twin snakes model was presented to detect an optimal seam line automatically in image mosaic. The color similarity and texture similarity between the images are formulated in the energy function of snakes. With this method an optimal seam line can be found automatically which is less differences in color and texture between the images. In the process of selection and iteration of the snakes the difference information of the images in color and texture is used. The seam line detected by the twin snakes model is more superior to that detected by normal method. Finally, a good result with this method was demonstrated in the experiment.