中文摘要：

针对原木图像存在年轮、锯痕以及原木之间紧密接触的问题，提出两阶段分割方法，即依据原木目标和背景的颜色差异应用K均值算法进行聚类得到相互紧密接触的原木目标分割结果；然后应用分水岭分割算法将粘连的原木目标加以分离．结果表明：与直接用分水岭算法和K均值算法相比，本方法能够有效地分割出原木目标．

英文摘要：

To solve the problems such as presence of tings, saw marks and close contact between logs, two stages segmentation method was proposed. According to the differences of color between logs and background, K-means algorithm was used to cluster and get the target of closely contacted logs, watershed algorithm was used to disjoint the target of closely contacted logs. The results revealed that this method could segment log targets more egiciently than watershed algorithm or K means algorithm.