Research On Ballistics Of Bullet Comparing System Based On Three Dimensional Surface Tonography

The comparison and verifying of the bullets trail could identify the fired guns toprove the criminal, the efficiency of comparison using microscope is vefy low, so it'sneed for an more efficient and more accurate method. At present there are manymethods based on computer image processing, but they are based on two-dimensionalgray image, due to the influence of ambient light, and not include the information ofdepth, the reliability of the results of compared grayscale image is not high enough, soit's also need lots of manual matching to distinguish the similarity of the two bullets,the efficiency of comoparison is rarely improved.In order to improve the efficiency of bullet trail comparison, the recent scientificstudies indicate that the depth topography of the bullet trail is very importantinfromation when we compared the bullets, so how to accurately obtain the depthtexture is emphases in research. On the basis of the measuring status in the traces ofthe bullets, according to the requirements of the system, at the same time to obtain asufficiently high measurement accuracy, nor will be destroyed the bullet textureinformation, we present optical non-contact measurement method. The analysis showsthat the grating projected method based on struct optical can get the objects highinformation by analyzing the change grating and there are several advantages, such asaccuracy of data is very high, the measurement speed is quick, gets the texture of theobject without damage et al..., all of these fit the requirements of bulletthree-dimensional texture information measurement. So we choose the gratingprojected phase method, then the importent thing is to measure the three dimensionalinformation of bullets, and analyzed the similarity of the bullets simultaneously.In the field of the structed light projection grating phase shift measurement, themeasured object are relatively large objects, the study is relatively less in themeasurement of microscopic field, analyzing the grating projected phase methodparticularly and considering the system of bullets, the paper presents the structure ofall measurement system. The most importent aspects of the projection grating phaseshift measurement is phase calculating and unwrapping, in the first place, we selectedadaptive Wiener filter and layered soft-threshold wavelet to denoising, proposed thesixteen phase shift based on average method to calculated the phase, and the accuracy is improved, in terms of phase unwrapping, introduted the details of the gray code,and describes its principle, analyzed its drawbacks. We proposed the gradient-orientedphase unwrapping algorithm based on multi-frequency, experimental results showsthat this algorithm could overcome the discontinuity point and the transition point ofphase unwrapping, the measurement accuracy is improved effctively.In the calibration of the measurement system, all calibration parameters of systemare described particularly, and calibrated the optical distortion of the system, weselected the high-precision Leica theodolite as auxiliary equipment, used partionedfitting algorithm to calibrated the overall distortion, utilized the calibration coefficientto corrected the phase, then carried out the calculation of phase-height mapping.Aim to bullet extraction and bullet comparison, first we wrote the software usedOpenGL, added the pseudo-color function; talked about the principle based on ellipsefitting to extract the bullet trace, and given the specific methods and formulas of thecoordinate transformation. For this method can't handle the case of deformation of thebullet, proposed the Gaussian filter midline to extracted the bullet trace, descriped thedesin process of the filter in detailed, given the analysis of the filter approximationperformance, reduced data redundancy, improved accuracy and efficiency of theextraction, discussed the extracted results in the different cut-off wavelength. Basedon the above extracted results, we used the cross-correlation algorithm to characterizethe degree of similarity of the two bullets. In order to increase the reliability of theresults, we proposed the data fusion of2D+3D for artifical assisted analysis.In order to solve the bullet topography information destructed while the bulletwas deformated seriously, we analyzed the cartridge case, using voxel-basedcross-correlation measure method to calculate the similarity. In order to overcome theincosistency of coordinates, proposed the iterative algorithm based on adaptivevariable coefficient for the calculation of the optimal similarity, given theexperimental procedure, the result shows that this method can improve the reliabilityof matching.Experimental results show that the algorithm and hardware design of this systemmeet the bullet comparison needs and it achieves the desired purpose.