A novel steganography algorithm based on Curvelet transform against JPEG attacks

Comparing with traditional multiscale transform, the Curvelet transform can find optimally sparse epresentations of objects, with discontinuities along C2 edges. Based on dividing blocks curvelet transform the paper present a novel steganography algorithm. Because the changes of the coefficients in detail scale layers keep energy conservation in the ridge domain, the secret message can be inserted through blocks energy quantization. The steganography algorithm is blind to extraction and robust to common attacks, especially to JPEG. The experimental results show that it can improve the secret message capacity, and suit to covert commutation against JPEG attack, which insert messages in detail1 and detail2 scale layers. It can obtain more strong robustness to semi-fragile watermarking which insert messages only in detail1 scale layers.