In general, the approximation of Discrete Cosine Transform (DCT) is used to decrease computational complexity without impacting its efficiency in coding. Many of the latest algorithms used in DCT approximation functions have only a smaller DCT length transform of which some are non-orthogonal. For computing DCT orthogonal approximation, a general recursive algorithm is used here, and its length is obtained using DCT pairs of length N/2 of N addition cost in input pre-processing. The recursive sparse matrix has been decomposed by using the vector symmetry from the DCT basis in order to achieve the proposed approximation algorithm that is highly scalable to enforce the highest lengths software and hardware by using a current 8-point approximation to obtain a DCT approximation with two-length power, N>8. © Published under licence by IOP Publishing Ltd.