高斯滤波器在子弹三维图像特征提取中的应用

为了弹痕的自动比对,建立了基于结构光的子弹三维纹理信息获取系统。由于在采集三维信息时子弹的中轴线与采集系统的y坐标轴很难做到完全重合,这需要进行相应的坐标旋转,尤其是在子弹变形时,无法利用子弹近似圆柱提取这一信息,这都给弹痕提取带来了很大的困难。本系统采用高斯逼近滤波器的方法获取高斯滤波中线,然后将三维弹痕信息与高斯滤波中线相减,进而提取出子弹的弹痕信息。利用冲激响应不变法设计了十二级的IIR型数字高斯滤波器,所设计的高斯逼近滤波器最大幅度偏差只有1.38%,大大优于ISO11562规定的2.5%的最大幅度偏差。此方法使得弹痕提取过程更简单,提取结果更精确,对后续弹痕比对的准确率有明显的提升。

Application of Gaussian Filter in Extracting Bullet Feature of 3-D Bullet Image

In order to compare the bullet automatically,the 3D texture information acquisition system of the bullet based on the structure light is established.Because the central axis and acquisition system axis y is difficult to completely overlap when we acquired three-dimensional information of bullets,so the corresponding coordinate rotation is need,especially when the bullet was deformed,it is unable to utilize the information that the bullet is approximate a cylindrical,this is very difficult for bullet extraction.Therefore this paper used the Gaussian approximation filter to get the mean line determined by Gauss filter,and then 3-D bullet information subtracts from mean line determined by Gauss filter,the bullet texture was extracted.Twelve order IIR digital Gauss filter is designed using the impulse response method,its amplitude transmission characteristic deviation is less than 1.38%,significantly better than the ISO11562 specified 2.5%.This method makes the bullet extraction process more simply,and the extraction result is more accurately,improving the accuracy of bullet comparison obviously.