基于纹理特征匹配的快速目标分割方法

Fast object segmentation based on texture matching

目标分割方法是工业自动化、在线产品检验、生产过程控制等领域的关键技术之一。基于特征匹配策略,研究了如何增强纹理特征的区分能力以及如何快速分割特定的目标。在纹理特征提取方面,首先通过形态学处理获取图像细节信息,然后对细节信息进行过滤进而得到鲁棒的局部信息,最后融合局部二值模式用于增强特征的区分能力。在目标分割方面,基于纹理特征和变化分解窗口框架,根据目标窗口和待定目标窗口间的特征距离分等级的变化分解窗口的尺度,从而快速定位到目标。最后,在纹理数据集上验证了方法的有效性。

Target segmentation is widely applied to automated areas. Based on feature matching strategy,we focus on extraction of discriminative feature and fast segmentation of target. To extract more discriminative feature,we propose to fuse detailed information of image obtained from morphological processing with local binary pattern. To efficiently locate the target,we propose to change the size of decomposing widow in a rated way,according to feature distance between the target and the candidate window. Experiments demonstrate that our method achieves better performance.