Identifying Critical Factors Influencing Quality of Blood Vessel Information in JPEG Compressed Skin Images

Recent research results demonstrate the potential of using blood vessel patterns for criminal and victim identification. With newly developed methods, blood vessel patterns under human skin are possible to be visualized from evidence images (e.g., Child sexual abuse images). However, these images are always taken by consumer cameras and compressed by the JPEG method, which can degrade the blood vessel information seriously. In this paper, an analysis is conducted to find out the critical factors influencing the quality of the blood vessel information. Simulations using different compression ratios are performed on skin images with different resolutions. The results indicate that the quality of blood vessel information is controlled by several low frequency discrete cosine transform (DCT) coefficients in the Y, U and V channels. The findings will be useful for designing algorithms to restore the blood vessel information lost in the JPEG compression process in future research.