Study On Shoe Prints Identification Algorithms Based On Texture And Shape Characteristics

In the investigation of criminal cases, footprints and shoe prints taken from crime scenes are important in serial and parallel processing of cases. At present, only a few police criminal technique departments in big or middle cities of our countries have set up shoe prints storeroom and computers are not used in the management of the crime scene shoe-print photos. The amount and varieties of shoe prints extracted from crime scenes by criminal investigation policemen are very huge. If not managed by scientific methods, the more shoe prints are accumulated, the more probably they may get spoiled or even lost. In addition, large number of shoe prints accumulated over a long period of time and managed by manual work, also faces the danger of being mistaken or confused. So it is necessary to find a kind of image processing method under this background, to make use of the characteristics of fast speed of computers, to carry out shoe prints identification and classification automatically, and to play an important role in criminal investigation.In this dissertation, the concrete analysis of the classification of shoe prints collected from shoe manufactories are carried out first. Then based on the feedback opinions of police operators and the research works in image processing field that have been done in recent years in Dalian Maritime University, the algorithms are brought forward as following:1. Improving the binarization algorithm which forms the base work of identification of complex textures.2. Making use of morphological filter combining with shape characteristics in order to identify dot textures. The interlaced textures are identified by detecting cross dots in the thinned edges of binary image.3. Making use of Canny method to detect image edges. A new algorithm with memory for image boundary tracing is proposed. After image boundary tracing, the classification of the traced edges is made.4. Adopting the chain code method to analyze the main edges of not-closed edges extracted from those which have been traced. According to the characteristics of lines and arcs, lines and arcs in line texture and line-wave-shape and arc-wave-shape in wave-shape texture are distinguished.5. For closed edges, the polygon textures and the circular textures are distinguished by making use of the circular degree and the ratio of the average value and the maximum value after the close area projecting. According to the feature of shoeprints, improvement on the Hough transform method is made in order to distinguish the ellipse and circle in circular textures.