基于邻域梯度向量的弹壳图像拼接算法研究

Neighborhood Gradient Vector BasedCartridge Case Image Mosaicing Algorithm

研究用于取证的弹壳图像拼接问题,提高拼接准确率。由于枪弹壳体形小,在多角度拍摄用于拼接的图像时,不可避免会重叠拍摄使得获取的图像间相似性较大,传统的基于特征匹配的弹壳图像拼接方法,易受图像间相似性影响而造成图像特征误匹配,导致弹壳图像拼接准确率不高的问题。为了解决这一难题,提出基于邻域梯度向量的弹壳图像拼接方法。通过计算图像分块邻域梯度向量,据向量距离得到初步匹配集,并采用自适应参数组合算法去除初步匹配集中的错误匹配,然后利用聚类算法完成弹壳图像的拼接融合,避免了传统方法受图像相似性的影响而造成拼接准确率不高的问题。结果表明,此种方法能够去除图像相似性的干扰,准确完成弹壳图像的拼接,取得了满意的结果。

Research the evidence cartridge case image mosaicing problems and improve mosaicing. Based on the neighborhood gradient vector mosaicing, the method calculates the neighborhood gradient vector of the cartridge case image blockd, and according to vector distances, preliminary matching set was obtained. Using adaptive parameter combination, the mistake matchings by preliminary matching algorithm were removed, and then the clustering algorithm was used to complete the mosaicing of image mosaicing. Simulation results show that this method can eliminate image similarity interference, accurately complete the mosaicing, and achieve satisfactory results.