Modified STDM based on hybrid transform and sub-block correlation

摘要

In this paper, two modified spread transform dither modulation (MSTDM) algorithms are proposed based on combination of transforms of discrete wavelet transform and discrete cosine transform (DD: abbreviation of the two discrete transform). They are called MSTDM-CO-DD algorithm based on correlation between adjacent blocks and MSTDM-PCO-DD based on the correlation between adjacent blocks after pretreatment of the adjacent sub-block before embedding. In both algorithms, we first make wavelet decomposition of the image, and divide the low-frequency sub-band into blocks. MSTDM-CO-DD gets the projection vectors and quantization steps from the previous sub-block to modulate the latter sub-block based on the correlation between adjacent blocks. As a result, MSTDM-CO-DD solves the problem of the difference of projection vector and quantization step between embedding and detecting. MSTDM-PCO-DD exchanges the coefficients of the previous sub-block and the latter one before embedding. As a result MSTDM-PCO-DD is more robust. The quantization step of the proposed algorithms varies with the brightness of the carrier image adaptively based on modified Watson model. From the results of numerical simulation, it is obvious that our proposed algorithms are robust to many attacks. Moreover, based on the Quantized Projection Methods our algorithms have been analyzed in performance.