Zeno Geradts

Methods for identification of images acquired with Digital cameras

From the court we were asked whether it is possible to determine if an image has been made with a specific digital camera. This question has to be answered in child pornography cases, where evidence is needed that a certain picture has been made with a specific camera. We have looked into different methods of examining the cameras to determine if a specific image has been made with a camera: defects in CCDs, file formats that are used, noise introduced by the pixel arrays and watermarking in images used by the camera manufacturer.

The image-database REBEZO for shoeprints with developments on automatic classification of shoe outsole designs

Abstract A database for footwear outsole designs has been developed on a PC. The database consists of three files: shoes of suspects, shoeprints from the scene of crime and shoes available from the shops. For REBEZO an algorithm is implemented for the automatic classification of outsole patterns. The algorithm first segments the shoeprofiles in different profiles. The Fourier-features are calculated for those profiles. The best Fourier features are selected and are classified with a neural network. By using this algorithm many different shapes can be recognised. Integrating the results of the invariant moments in the neural network will give better results.

Using the ENF Criterion for Determining the Time of Recording of Short Digital Audio Recordings

The Electric Network Frequency (ENF) Criterion is a recently developed forensic technique for determining the time of recording of digital audio recordings, by matching the ENF pattern from a questioned recording with an ENF pattern database. In this paper we discuss its inherent limitations in the case of short --- i.e., less than 10 minutes in duration --- digital audio recordings. We also present a matching procedure based on the correlation coefficient, as a more robust alternative to squared error matching.

Source Camera Identification for Heavily JPEG Compressed Low Resolution Still Images

Abstract:  In this research, we examined whether fixed pattern noise or more specifically Photo Response Non‐Uniformity (PRNU) can be used to identify the source camera of heavily JPEG compressed digital photographs of resolution 640 × 480 pixels. We extracted PRNU patterns from both reference and questioned images using a two‐dimensional Gaussian filter and compared these patterns by calculating the correlation coefficient between them. Both the closed and open‐set problems were addressed, leading the problems in the closed set to high accuracies for 83% for single images and 100% for around 20 simultaneously identified questioned images. The correct source camera was chosen from a set of 38 cameras of four different types. For the open‐set problem, decision levels were obtained for several numbers of simultaneously identified questioned images. The corresponding false rejection rates were unsatisfactory for single images but improved for simultaneous identification of multiple images.

Source video camera identification for multiply compressed videos originating from YouTube

The Photo Response Non-Uniformity is a unique sensor noise pattern that is present in each image or video acquired with a digital camera. In this work a wavelet-based technique used to extract these patterns from digital images is applied to compressed low resolution videos originating mainly from webcams. After recording these videos with a variety of codec and resolution settings, the videos were uploaded to YouTube, a popular internet video sharing website. By comparing the average pattern extracted from these resulting downloaded videos with the average pattern obtained from multiple reference cameras of the same brand and type, it was attempted to identify the source camera. This may be of interest in cases of child abuse or child pornography. Depending on the codec, quality settings and recording resolution, very satisfactory results were obtained.

Content based information retrieval in forensic image databases.

This paper gives an overview of the various available image databases and ways of searching these databases on image contents. The developments in research groups of searching in image databases is evaluated and compared with the forensic databases that exist. Forensic image databases of fingerprints, faces, shoeprints, handwriting, cartridge cases, drugs tablets, and tool marks are described. The developments in these fields appear to be valuable for forensic databases, especially that of the framework in MPEG-7, where the searching in image databases is standardized. In the future, the combination of the databases (also DNA-databases) and possibilities to combine these can result in stronger forensic evidence.

Source Camera Identification for Low Resolution Heavily Compressed Images

In this paper, we propose a method to exploit photo response non-uniformity (PRNU) to identify the source camera of heavily JPEG compressed digital photographs of resolution 640 times 480 pixels. Similarly to research reported previously, we extract the PRNU patterns from both reference and questioned images using a two-dimensional high-pass filter and compare these patterns by calculating the correlation coefficient between them. To deal with the low quality compressed image material, we propose a simple and effective way to obtain the PRNU. We did extensive experiments for both the closed and open set source camera identification problem with a set of 38 cameras of four different types. For the closed set problem accuracies as high as 83% for single images and 100% for around 20 simultaneously identified questioned images were obtained. For the open set problem, decision levels were obtained for several numbers of simultaneously identified questioned images. The corresponding false rejection rates were unsatisfactory for single images, but improved substantially for simultaneous identification of multiple questioned images.

A New Approach to Automatic Comparison of Striation Marks

A database for toolmarks (named TRAX) has been developed on a PC with Microsoft Windows. The database is filled with video-images and administrative data about the toolmarks (width, kind of toolmark, etc.). A comparison screen in TRAX makes it possible to compare images of toolmarks. The system works with the Screen Machine multi media-board. A new algorithm for the automatic comparison of digitized striation patterns has been developed. The algorithm works well for deep and complete striation marks and will be implemented in TRAX.

Using anisotropic diffusion for efficient extraction of sensor noise in camera identification.

Abstract:  Each digital camera has an intrinsic fingerprint that is unique to each camera. This device fingerprint can be extracted from an image and can be compared with a reference device fingerprint to determine the device origin. The complexity of the filters proposed to accomplish this is increasing. In this note, we use a relatively simple algorithm to extract the sensor noise from images. It has the advantages of being easy to implement and parallelize, and working faster than the wavelet filter that is common for this application. In addition, we compare the performance with a simple median filter and assess whether a previously proposed fingerprint enhancement technique improves results. Experiments are performed on approximately 7500 images originating from 69 cameras, and the results are compared with this often used wavelet filter. Despite the simplicity of the proposed method, the performance exceeds the common wavelet filter and reduces the time needed for the extraction.

Pattern recognition in a database of cartridge cases

Several systems exist for collecting spent ammunition for forensic investigation. These databases store images of cartridge cases and the marks on them. The research in this paper is focused on the different methods of feature selection and pattern recognition that can be used for comparison. For automatic comparison of these images it is necessary to extract firstly the useful parts of the images. On databases of 3800 images several processing steps have been tested and compared. The results and methods, which have been implemented, are presented. The usual correlation methods based on gray values of all relevant image data have been tested. They were useful in the database. Further invariant image descriptors and the a trous wavelet transform have been implemented. These methods are promising, however more investigation is needed for the use of these methods.