基于余弦距离的智能拼图算法的应用研究

Application of Jigsaw Puzzles Based on Cosine distance

由于智能拼图算法根据相邻块的相似度进行组装,如何计算拼图块之间的相似度度量至关重要。目前,大多数拼图算法侧重于单纯地计算碎片边缘相似性,对碎片内容相似度与边缘相似度结合的算法研究甚少。因此,本文提出一种计算两个拼图块之间的余弦距离的内容相似度评分的算法,研究内容相似度对智能拼图的重要性,并对部分网络图片以及笔者拍摄的图片进行实验。实验结果表明,本文算法具有更高的准确率。

Because the intelligent puzzle algorithm is assembled according to the similarity of adjacent blocks, how to calculate the similarity measurement between puzzle blocks is very important. At present, most jigsaw puzzle algorithms focus on the simple calculation of fragment edge similarity, and there is little research on the algorithm combining fragment content similarity and edge similarity. Therefore,this paper proposes a content similarity scoring algorithm to calculate the cosine distance between two puzzle blocks, and studies the importance of content similarity to intelligent puzzle. The experimental results show that the algorithm has higher accuracy.