Security Research Centre, School of Computer and Security Science, Edith Cowan University
Place of Publication
Perth, Western Australia
David Michael Cook
Ballistics firearm identification based on image processing is of paramount importance in criminal investigation. The efficiency of traditional ballistics imaging system is heavily dependent upon the expertise and experience of end-user. An intelligent ballistics imaging system is highly demanded to address this issues. The segmentation of cartridge case object from the original image is a key step to subsequent process. It is very difficult to segment cartridge case from the original image using traditional threshold based methods due to the shadows or unlimited environments to acquire image. In this paper, we proposed a novel approach based upon the colour invariant and geometrical shape of objects. The experimental results show the proposed method can precisely segment the objects in various images such as heavy shadow, low contrast, uneven illumination etc.