基于加强模糊聚类的航空行李图像超像素分割

在自助行李托运系统拍摄的行李图像库中检索错误运输的行李时，图像背景会严重影响检索精度。针对该问题提出了一种加强模糊聚类算法（EnFCM）的超像素分割方法，实现了行李目标区域的提取。通过多尺度形态学重建梯度图像，设计了自适应上限尺度的分水岭超像素预分割算法，得到多个独立的超像素区域。对超像素图像进行直方图统计，并结合分水岭分割参数和实际行李图像内容的类别数量进行超像素的加强模糊聚类，得到行李区域。通过多个实际行李图像的分割实验验证了算法的有效性，平均分割精度达到93%，超过多个典型的分割算法。

Super-pixel segmentation of air baggage image based on enhanced fuzzy clustering

In the baggage image database captured by the self-service baggage check-in system, the image background seriously affects the retrieval accuracy of wrongly transported baggage. To solve this problem, a super-pixel segmentation method based on Enhanced Fuzzy C-Means Clustering Method (EnFCM) is proposed to extract the target baggage area. Through multi-scale morphological reconstruction of gradient images, an adaptive upper scale watershed super-pixel pre-segmentation algorithm is designed to obtain multiple independent super-pixel regions. Based on the histogram statistics of the super-pixel image, combined with the watershed segmentation parameters and the number of categories of the actual baggage image content, the EnFCM segmentation is conducted to extract the baggage area. The segmentation experiments of several real baggage images verify the effectiveness of the algorithm with the average segmentation accuracy reaching 93%, surpassing that of several typical segmentation algorithms.