摘要
在拍摄光滑客体表面的无色指纹时,一般采用定向反射光照相的方法,但是对于光滑纸板等表面,潜在指纹与背景之间的反射亮度差很小,传统方法难以进行拍摄提取。使用不同方向的光源配光,每一方向光源采集一幅图像,然后使用小波变换,将图像分解为不同尺度空间上的一系列子图像。对于低频子图像,采用取平均的融合规则;对于高频子图像,采用基于窗口的区域能量决策进行系数融合。最后对融合后的图像进行基于自适应引导滤波的多尺度Retinex图像增强,以增加指纹图像对比度。实验结果证明,该算法可以解决光滑物面指纹图像难以提取的问题,从而得到较为清晰的指纹图像。
When taking a colorless fingerprint on the surface of a smooth object, we usually adopt the method of directional reflec ted light photography. However, for glossy cardboard, the difference in reflection brightness between the potential fingerprint and the background is small, and the traditional method is difficult to extract. We use different directions of the light distribu tion, to capture an image at each direction of the light source. Then by using wavelet transform, the image is decomposed into a series of sub images in different scales. For low frequency sub images, the average fusion rule is adopted. For the high fre quency sub images, window based regional energy weighting is used to perform coefficient fusion. Finally, the multi scale Ret inex image enhancement with adaptive guided filtering of the fused image is used to increase the contrast of the fingerprint im age. Experimental results show that this algorithm can solve the problem of that it is difficult to extract the smooth object sur face fingerprint image so that a clear fingerprint image can be ottained.
作者
祝念
高思莉
蔡能斌
ZHU Nian;GAO Si-li;CAI Neng-bin(Shanghai Institute of Technical Physics of the Chinese Academy of Sciences;CAS Key Laboratory of Infrared System Detection and Imaging Technology,Shanghai 200083,China;Shanghai Key Laboratory of Criminal Scene Evidence,Institute of Forensic Science,Shanghai Public Security Bureau,Shanghai 200042,China;University of Chinese Academy of Sciences,Beijing 100049,China)
出处
《软件导刊》
2018年第10期157-160,共4页
Software Guide
基金
国家"十三五"国防预研项目(Jzx2016-0404/Y72-2)
上海市现场物证重点实验室基金项目(2017xcwzk08)
中国科学院青年创新促进会项目(2014216)
