Steganography is art of hiding a secret media in another media. One of the major challenges in steganography is robustness, since the stego-signal need to survive multiple kinds of data processing. Low pass filtering, for example JPEG compression, is known as a common attack against stego-signal in image based steganography. In this paper, we present a new method of image steganography that is extremely robust against JPEG compression while allowing error free information extracting. The method is based on 2D lossless wavelet transform and convolution error correction coding. Experimental results show that hidden information can be retrieved with zero bit error rates even when the stego-image experienced maximum JPEG compression.