For improving the lossly image coding efficiency of integer wavelet transform (IWT) based on lifting scheme, the scaling factor is used in the lifting steps. However, the scaling factor is often an irrational number, which requires three additional lifting steps to assure the integer character of transform results. They not only increase computational complexity of integer wavelet transform, but also raise the effect of rounding error on image compression. This paper presents a new image coding algorithm using optimal scaling scheme and quadtree partitioning. The new algorithm decomposes the image using integer wavelet without scaling factor and then only performs a multiplication on every subband with the new scaling factor, which reduces the computational complexity of the integer wavelet transform. During the image encoding, a new quadtree partitioning scheme is used to improve the searching efficiency of the significant coefficients. The experimental results show that the new coding algorithm has not only the well lossly compression efficiency, but also the lower computational complexity than the common IWT. This new method is valuable for future remote sensing and medical image compression.