基于图像识别的橡胶鞋底Schallamach波自动计数方法  

Automatic Counting of Schallamach Waves from Rubber Shoes’Outsoles Based on Image Recognition

作　　者：全永志 于奎栋 刘珞晰 QUAN Yongzhi;YU Kuidong;LIU Luoxi(Shanghai Police College,Shanghai 200137,China;Shanghai University of Political Science and Law,Shanghai 201701,China)

机构地区：[1]上海公安学院,上海200137 [2]上海政法学院,上海201701

出　　处：《刑事技术》2023年第2期182-188,共7页Forensic Science and Technology

基　　金：新入校青年教师资助计划(C-02-21-004-53)

摘　　要：在犯罪现场中,鞋印是一种常见的痕迹物证。Schallamach波是橡胶鞋底上的一种磨损痕迹特征,对鞋印的检验鉴定具有一定的实际意义。目前对Schallamach波纹线采用人工统计,耗时耗力,因此本文利用图像识别提出一种自动计数方法。首先对图像进行二值化处理实现纹线与背景的初步分离,然后对图像进行形态学处理,利用迭代腐蚀膨胀操作分离粘合纹线,消除杂点,填充孔洞。通过Zhang-Suen(ZS)细化操作提取纹线骨架,在剔除干扰纹线后,每隔10个像素进行一次计数,得到的平均数作为全图纹线数目。使用不同密度图像进行验证,结果表明,在低、中密度时误差均控制在1条左右,计数时间均为1 s左右。本文方法能够对Schallamach纹线高效计数,与人工计数相比耗时显著降低,在面对大范围统计时将节约大量时间成本。Footwear impression,one common kind of evidence collected at crime scene,is usually identified through its acquired characteristics.With the development of technologies and ever-upgrading devices used for crime scene investigation,the related public security professionals are competent of finding clearer features of the acquired characteristics on footwear impressions.Among the characteristics,the Schallamach pattern,capable of revealing the specificity and reproducibility of footwear impression,shows wave-shaped wearing features on rubber shoes’outsoles,accordingly demonstrating practical significance for forensic footwear impression examination and solution of relevant criminal cases.Uninterestingly,Schallamach waves are at present mostly counted by manual counting,a time-consuming and labor-intensive task,hence very likely resulting in erroneous statistaical conuts due to human unintentional factors.Therefore,an automatic counting approach was proposed here on the basis of image recognition.Firstly,the footwear impression image was binarized to achieve the preliminary separation of the ridges and the background.Then,the image was morphologically processed,thereby having the bonded ridges detached via iterative erosion and expansion operation so that the isolated noising dots were to eliminate and the holes fill wherever presence of incomplete printing.For convenience of subsequent counting,the ridge skeleton was extracted with Zhang-Suen thinning manipulation.In order to count the whole image evenly and candidly,those disturbing ridges were removed in the vertical direction and counting numbers were taken with every other 10 pixels,thereafter having an average number obtained to use as the number of ridges in the whole image.The approach was successively verified into images of different densities,having its error occurrence controlled at about 1 for low and medium densities and the counting time about 1 second.The approach adopted here can efficiently count Schallamach waves,indeed spending significantly le

关 键 词：鞋底磨损特征 Schallamach波 图像识别 自动计数 

分 类 号：DF794.1[政治法律—诉讼法学]