Abstract: Stitching and restoration method of the image fragments has important applications in the judicial evidence reconstruction and military intelligence acquisition and other fields. Firstly a calculation model on integrated similarity is designed according to the features of the image fragments. In order to improve the accuracy of a stitching, a “double-breasted” stitching method is designed. Secondly according to the greedy principle, an effective stitching algorithm of the image fragments is designed through expansion and around the back, and according to the reconstruction features of the image fragments, the guidelines for manual intervention are developed to achieve a visual reconstruction of the image fragments under human-computer interface system. Finally, the effectiveness of the method is verified by experiments and the stitching problems of cross-cut shredded image fragments are solved effectively.

Key words: image fragments, integrated similarity, human-computer interaction, double-breasted, greedy algorithm

摘要： 图像碎纸片的拼接与复原在司法物证复原和军事情报获取等领域有着重要的应用。根据图像碎片的特点建立了集成相似度的计算模型，设计了一种“双排扣”拼接方法来提高一次拼接的准确性；根据贪心准则设计了一种向四周扩展和绕回的图像碎纸片有效拼接算法，并根据图像碎纸片拼接特点制定了人工干预准则，实现了可视化的人机交互拼接系统。最后，通过实验验证了该方法在解决图像碎纸片拼接问题上的有效性。

关键词: 图像碎片, 集成相似度, 人机交互, 双排扣, 贪心算法