Method for detecting digital image integrity violations due to its block processing
Kobozeva A.A., Іван Ігорович Бобок · Radiotekhnika · 2019
The detection of data integrity violations is one of the main issues for information security professionals. In particular, this problem is relevant for digital images (DI), which this work is dedicated to. Block processing of DI has become typical for modern steganographic algorithms, and it is widely used today. Block processing is any processing that involves saving the resulting image in lossy formats (JPEG, JPEG2000). It is local image processing that occurs during cloning and photo editing. In this regard, the aim of the work is to provide effective detection of the DI integrity violations due to its block processing, regardless of its specific type by developing an appropriate method. The method is based on the results obtained by the author previously through research studies of the formal parameters of DI matrix blocks. The proposed method and its algorithmic implementation (SA) have no analogues, that is confirmed by data obtained from open sources since existing methods for detecting image integrity violations are usually designed for specific perturbations. The paper presents the results of a computational experiment, which has confirmed the high efficiency of the developed algorithm in the conditions of all considered perturbations, including block processing. Comparison of the SA effectiveness with analogues was carried out under conditions of specific perturbations. In all cases considered, the efficiency of the developed algorithm is comparable with the best of modern analogues. When using the SA as a steganalysis algorithm for detecting the LSB-embeddings in containers saved with losses, it remains effective under conditions of low capacity of the hidden communication channel (0,01 bit/pixel), while most of analogues algorithms are inoperative.