Abstract: A method for fast transform coding in image compression is provided. The method includes performing prediction with respect to a current block, calculating a prediction error with respect to the current block, performing transform and quantization with respect to the current block, performing dequantization and inverse transform with respect to a transform coefficient generated through the transform and quantization, generating a reconstruction signal by summing a prediction signal generated in the performing of the prediction and a residual signal generated in the performing of the dequantization and inverse transform, calculating a rate-distortion (RD) cost related to the reconstruction signal, determining whether a size of the current block is a minimum block size, partitioning the current block into a plurality of sub-blocks, and performing transform coding with respect to the sub-blocks in response to determination that the size of the current block is not the minimum block size.