因为在实际网络环境下的隐蔽通信检测系统中，获得的图像信源往往包含多种多

Because in the covert communication detection system in the actual network environment, the obtained image source often contains a variety of image content, which is obviously different from the research of the steganalysis algorithm of a single image content in the laboratory environment, which will cause The general blind detection classifier has the problem of "mismatch" in the statistical characteristics of the carrier image source in the training phase and the test phase, thereby reducing the detection effect of the detection algorithm, which brings difficulties to the hidden information detection in the network heterogeneous image environment.

Because in the covert communication detection system in the actual network environment, the obtained image source often contains a variety of image content, which is obviously different from the steganalysis algorithm research of single image content in the laboratory environment, which will cause the "mismatch" of the statistical characteristics of the carrier image source of the general blind detection classifier in the training stage and the testing stage, Thus, the detection effect of the detection algorithm is reduced, which brings difficulties to the hidden information detection in the network heterogeneous image environment.

Because in the actual covert communication detection system under the network environment, the obtained image sources often contain various image contents, which is obviously different from the research of steganalysis algorithm with single image content in the laboratory environment, and will cause the "mismatch" problem of the statistical characteristics of the carrier image sources in the training stage and the testing stage of the general blind detection classifier, thus reducing the detection effect of the detection algorithm, which brings difficulties to the detection of hidden information under the network heterogeneous image environment.