Image hiding is hiding a secret image embedding to a cover image. This paper intends to propose an image hiding method. First, a secret image is compressed by BST. Meanwhile, dividing the compressed data into different ranks depends on the significance. Then, the cover image is transformed into frequency domain data by discrete wavelet transformation. The dominant part of compressed data is hidden to the lower frequency sub-band which isn’t easy to destroy; then hide the trivial data into the higher frequency sub-band. Finally, when the dominant part of the compressed secret image data is damaged, the error data will be modified by the hamming code method. In addition, when the trivial data of the compressed data of the secret image is damaged, it would affect the data relating to the block only. The proposed method has 2.25 bits/pixels of the hiding capacity. The image quality of stego-image and extracted secret image are measured in terms of peak-signal-to-noise ratio and discusses robustness of noise and compression.