Precise Location and Registration of Striation by Integrating Image Context and Second Derivative Edges Fusion

Precise location of striation in the image is a key technology to realize automatic examination and comparison for criminal striation tool mark. At present, the edge detection methods are often applied to extract lines from tool mark. However, the edge detection methods are sensitive to image noise, and easily produce false edges. The realistic striations are difficult to be located precisely and extracted by using edge detection methods for tool mark. The precise location approach of striation by integrating image context and second derivative edge fusion was proposed in the article. Firstly, the comparison line was defined based on the method of artificial identification. The second derivative of image was computed along comparison line. The ridge and groove regions of striation mark were calculated by utilizing rules fusion for derivative values according to the properties of second derivative edge. Then, the ridge and groove regions were optimized and normalized, false striations in the deep grooves were deleted as well, based on the experience of artificial identification. Lastly, the significant points among ridge regions were computed by integrating image context and the striations were located precisely. On the basis of striation locating precisely, the registration of experimental samples was per-formed using registration vectors such as ridges with second derivative maximal absolute values, ridges with maximal gray scale, deepest grooves, as well as distances among them. The experimental results show that the approach presented in the paper can locate the striations and make image registration precisely and fast. Higher rate of registration can be reached especially for striation marks produced by tools with prominent individual features on the work side. Obviously, the proposed method is helpful to establish automatic comparison system for striation tool mark.