ALGORITHM DESIGN FOR APPROXIMATION FOR BOTH HARDWARE AND SOFTWARE LEVEL

This paper presents a generalized recursive formula to obtain orthogonal approximation of DCT where roughly DCT of length might be produced from some DCTs of length at the cost of additions for input preprocessing. Many of the existing computations for approximation within the DCT target just the DCT of small transform measures and most of them are non-orthogonal. Approximation of discrete cosine transform (DCT) is helpful for reducing its computational complexity without major effect on its coding performance. We perform recursive sparse matrix decomposition and use the symmetries of DCT basis vectors for deriving the suggested approximation formula. Suggested formula is extremely scalable for hardware furthermore to software implementation of DCT of greater measures, plus it will make technique existing approximation of 8-point DCT to obtain approximate DCT connected getting an electrical two length, .We show the suggested approximation of DCT provides comparable or better image and video compression performance compared to existing approximation techniques. It's proven that suggested formula involves lower arithmetic complexity in comparison while using the other existing approximation computations. We've presented an entirely scalable reconfigurable parallel architecture for the computation of approximate DCT while using suggested formula. One distinctively interesting feature within the suggested design is it might be configured for the computation in the 32-point DCT or parallel computation of two 16-point DCTs or four 8-point DCTs obtaining a marginal control overhead. The suggested architecture can be found to provide several positive aspects in relation to hardware complexity, regularity and modularity. Experimental results acquired from FPGA implementation show the benefit of the suggested method