Research On Algorithm Of Visualization In 3-D Bullet Marks

In recent years gun-related cases have occurred frequently,and still seriously threaten the lives of citizens.The public security organ faces severe challenges in disposing these evidence of gun-related.One of the challenges is comparison and recognition of bullet marks,it is very important technique in the Criminal investigation.In comparison and recognition of bullet marks,the primary procedure is the visualization of bullet marks' data,then the comparison in uncharted bullet can be accomplished based on the visual graphic.After that the high similarity level bullet marks will be recognized,locating the homologous gun's information and acquiring the pivotal clue to detect these cases.The visualization of bullet marks graphic data is the basis of everything in process.It includes the study of 3D bullet marks graphic in several aspects such as: preprocess,feature extraction,three-dimensional reconstruction,render and so on.The visualization of bullet marks graphic data can realize the display of the depth information in the bullet marks feature and the observation,comparison of infinite free degree.The visualization can make the information more integrity in display the data of bullet marks and avert the interference by uneven lighting.Therefore the technique of the bullet marks' visualization has a pivotal impact in the recognition of bullet marks.Researchers have taken more and more attention on the studying of recognition system and the technique and algorithm of recognition has developed in recent years,nevertheless the aspect in visualization lacks research,and the display of bullet marks feature information is not robust.Especially in three-dimensional display the bullet marks feature information displayed fuzzily,marks cannot be recognized and the image quality is very poor.All in all the effect of bullet marks display immediately influences the accuracy of recognition.The visualization technology lags behind relatively,result in that the accuracy difficultly improve.Therefore in this paper we conduct a study of the technology and algorithm in visualization in 3-D data of bullet marks,in order to realize the visualization of the information of feature information in bullet marks,lay a foundation for the study of recognition,improve the accuracy recognition,detect the gun-related cases and strike crimes.The domestic identifying and comparing of bullet marks is mostly based on the 2-D gray scale image.And this has big limitation.The image cannot be moved in random directions,cannot be compared parameters information of the relative depth,width and angle.And on account of uneven illumination the ridge information of image cannot be displayed unbroken between the peak and valley.The three-dimensional bullet marks model can provide a boundless visual liberty,compared with the 2-D image it can be moved and rotated in random directions in space,at the same time it can keep a consistent look and a natural viewing experience.In the 3-D feature image comparing the researcher can observe and compare the information in different directions.So the study of 3-D visualization bullet marks in this paper is extremely rewarding.Objects what we can see in life all are three-dimensional and our eyes are accustomed to accept three-dimensional image information.In the future,the technology of image display will be progressing to high dimension and as a new technology,the visualization of 3-D Surface appearance will be widely used for the study of comparison and recognition.In this paper,the characteristics of bullet marks are introduced,the principles and processes of the three-dimensional data acquisition bullet marks are described.By analyzed the 3D bullet marks data,the data is transformed to make it more convenient for computer processing,and the preliminary display of the 3D bullet image is realized.Based on the preliminary display of 3D bullet marks image,the method of removing of outliers and noise is studied,and the filtering method is improved and compared with the traditional filtering method.The 3D bullet marks image noise removal,contour extraction and flattening are realized.In the three-dimensional feature extraction process,the feature of the bullet marks is extracted by using the algorithm which is proposed in this paper based on the slope change.And according to the algorithm,the comparison of the relative information of depth,width and angle parameters of 3D bullet marks is realized.The correct recognition rate is increased from 85.6% to 90.2% under the existing sample condition.As a result of decreased the process of the bullet slice position by Artificial selection,the processing time of a single set of bullet data cost from 120 s decreased to 32.7s.After the feature extraction of the bullet marks,the 3D reconstruction of the bullet marks image is carried out.The Delaunay triangulation algorithm is studied,and the Delaunay triangular mesh reconstruction of the 3D bullet image is realized,and illumination and gloss rendering are added.In the final stage,three-dimensional visualization of the unbroken bullet marks feature image is realized by registration and stitching of three-dimensional bullet marks image.