Forensic Camera Identification in Social Networks via Camera Fingerprint

Image-related crimes cause the urgent demand for tracing the origin of digital images. The breakthrough is a passive detection method via photo response non-uniformity (PRNU) analysis proposed by Lukáš et al. Recently, digital images are often shot with handheld devices (such as smartphones) and transmitted using social media (such as LINE). Most of the images are distorted (such as compressed and resized) during transmission. Previous studies are less focused on the impact of transmission compression through social networks. Thirty-one different Apple mobile phones were used to capture digital images in the experiment. Images were uploaded to the photo album via LINE software and then downloaded. The modified signed peak correlation energy (MSPCE) statistics is used to evaluate the correlation between the PRNU values of the disputed images and the pattern noise of the experimental devices. Experimental results show that the PRNU analysis method can effectively trace the source of the shot device using the distorted images which are compressed and resized during the transmission in LINE.

On the Relationship between Gravity Waves and Tropopause Height and Temperature over the Globe Revealed by COSMIC Radio Occultation Measurements

In this study, the relationship between gravity wave (GW) potential energy (Ep) and the tropopause height and temperature over the globe was investigated using COSMIC radio occultation (RO) dry temperature profiles during September 2006 to May 2013. The monthly means of GW Ep with a vertical resolution of 1 km and tropopause parameters were calculated for each 5° × 5° longitude-latitude grid. The correlation coefficients between Ep values at different altitudes and the tropopause height and temperature were calculated accordingly in each grid. It was found that at middle and high latitudes, GW Ep over the altitude range from lapse rate tropopause (LRT) to several km above had a significantly positive/negative correlation with LRT height (LRT-H)/ LRT temperature (LRT-T) and the peak correlation coefficients were determined over the altitudes of 10–14 km with distinct zonal distribution characteristics. While in the tropics, the distributions of the statistically significant correlation coefficients between GW Ep and LRT/cold point tropopause (CPT) parameters were dispersive and the peak correlation were are calculated over the altitudes of 14–38 km. At middle and high latitudes, the temporal variations of the monthly means and the monthly anomalies of the LRT parameters and GW Ep over the altitude of 13 km showed that LRT-H/LRT-T increases/decreases with the increase of Ep, which indicates that LRT was lifted and became cooler when GWs propagated from the troposphere to the stratosphere. In the tropical regions, statistically significant positive/negative correlations exist between GW Ep over the altitude of 17–19 km and LRT-H/LRT-T where deep convections occur and on the other hand, strong correlations exist between convections and the tropopause parameters in most seasons, which indicates that low and cold tropopause appears in deep convection regions. Thus, in the tropics, both deep convections and GWs excited accordingly have impacts on the tropopause structure.

The Theoretical Limits of Watermark Spread Spectrum Sequence

At present, the spread spectrum (SS) sequences used in watermark include i.i.d. random sequences and the sequences used in SS communications. They appear earlier than digital watermark. Almost no researchers pay attention to whether they are really fit for watermark. In this paper, we compare the SS watermark channel and the traditional SS communication channel. We find out that their correlation property is different. Considering cropping and translation attacks, we define watermark auto- and cross-correlation and propose Loose Autocorrelation and Tight Cross-Correlation (LAC&TCC) properties for SS watermark. The LAC&TCC properties are that, whether or not synchronized, the autocorrelation is equal or close to 1 and the cross-correlation is equal or close to 0. Accordingly, the peak correlation is divided into the peak autocorrelationRa(l)and the peak cross-correlationRc(l). We establish the lower bound ofRc(l)and the higher bound ofRa(l), respectively. The two bounds indicate that, no matter how small the cover is reserved, the extractor can always find a threshold to distinguish auto- and cross-correlation in theory.