A novel methodology that improves the computation rate of the 2-D discrete cosine transform (DCT) is proposed in this paper. By applying a new image decomposition scheme the image slice representation (ISR) followed by the image block representation (IBR) method, any gray-scale image can be described by a set of pixel blocks having certain intensity values. Based on this image representation, the computation of the 2-D DCT coefficients, useful for many image processing tasks, can be increased significantly. Appropriate experiments highlight the main advantages and establish the usefulness of the introduced method.