密度聚类法在图像分割中的应用

提出了使用密度聚类法解决图像分割的新思路。首先把数字图像按照点的分布情况建立图像样本数据库，然后利用基于密度聚类法的DBSCAN算法进行图像分割。该算法能找到图像样本比较密集的部分，概括出图像样本相对比较集中的类，并可在带有“噪声”的图像中进行聚类，完成图像分割。文章还针对DBSCAN算法的缺点，提出了DBSCAN算法的改进思路。

Application of Density Clustering in Image Segmentation

The new way of using density clustering to solve image segmentation problem is proposed. At first the image sample database is created according to the pixel distribution of digital images, and then DBSCAN algorithm based on density clustering method is used to do image segmentation. The algorithm can find the comparatively dense regions and sum up the relatively concentrated clusters in image sample, and it can cluster to complete image segmentation in the image with noises. Also for the disadvantages of DBSCAN algorithm the improvement idea is put forward.