中文摘要：

针对FH算法（Felzenszwalb和Huttenloch提出的图像分割算法）中存在的欠合并现象,在L＊u＊v＊彩色空间,结合Mean Shift算法,提出了一种改进FH图像分割方法,即IFH（Improved FH）.该方法首先采用Mean Shift算法获得均值漂移图像,再将图像由RGB空间转换到L＊u＊v＊颜色空间;然后,结合L＊u＊v＊彩色空间,采用FH算法构造带权无向图,基于图像的颜色特征进行分割.实验证明,与原算法相比,该方法在分割精度与分割质量上有了很大程度的提高.

英文摘要：

For the FH algorithm （Felzenszwalb and Huttenloch proposed image segmentation algorithm） which exists under combined phenomenon, in theL＊ u ＊ v ＊ color space,combined with Mean Shift algorithm, an improved FH image segmentation method is proposed, denoted as IFH （Improved FH）. Firstly, the method uses Mean Shift algorithm to obtain the mean shift image, then the image is converted from RGB to L＊ u ＊ v ＊ color space. Sec- ondly, combined with the L＊ u ＊ v ＊ color space and constructed weighted undirected graph using the FH algorithm, segmentation is realized based on im- age color features. Experiments show that, compared with the original algorithm, this method has greatly improved in the accuracy of segmentation and segmentation quality.