978-1-5386-1556-0/17/$31.00 ©2017 IEEE Requirement Based Transform Coefficient Coding Architecture for DCT/DST for HEVC Zahra Rauf Saleemi Faculty of Engineering and Computer Sciences, National University of Modern Languages, Islamabad, Pakistan zaleem@numl.edu.pk Gulistan Raja Faculty of Electrical and Electronics Engineering, University of Engineering and Technology, Taxila, Pakistan gulistan.raja@uettaxila.edu.pk Abstract— This paper proposes a Transform Coefficient Coding (TCC) architecture for High Efficiency Video Coding (HEVC) by using Selective Implementation Patterns. The architecture developed targets to exploit the benefits of trade- offs between number of cycles and complexity depending upon the requirements, hence reducing the execution time. The input configuration to be sent in parallel, semi parallel and serial manner, efficiently implements the design. The so- called method of Pipelining is the key to selection of different pixel combinations of input video signal, allowing to give maximum flexibility to compute the integral part of TCC, the One Dimensional Discrete Cosine Transform (1D-DCT) and Two-Dimensional Discrete Cosine Transform (2D-DCT). Furthermore, with a slight modification, the same design can be used for computing Discrete Sine Transform (DST) as well. Coding in Verilog HDL and Implementation in Spartan 3 FPGA Kit enhances the adaptability of the proposed scheme. By evaluating the best available option, lowest hardware complexity, with 11 Look Up Tables (LUTs) and 7 registers, is achieved. The idea is to introduce selectivity and reuse in architecture in Video Coding technique so that the hardware cost and computational time can be reduced. Keywords— HEVC; Transform coefficient coding; DCT (Discrete Cosine Transform); Clock cycles; Hardware architecture; Pipelining. I. INTRODUCTION The latest video coding standard employed now-a-days is High Efficiency Video Coding (HEVC). It was developed by Joint Collaboration Team on Video Coding (JCT-VC) of ITU-T SG16 WP3 and ISO/IEC JTC1/SC29/WG11. Proposals were called for the new standard in January 2010, and the initial responses were analyzed and reviewed by JCT-VC in its first meeting in April 2010. The first test model (HM1.0) was made available in October 2010 based on the optimal performance proposals. Several modifications have been made afterwards [1]. HEVC aims to achieve twice the compressional efficiency of its predecessor H.264/AVC without any detectable loss in the video quality with at least twice the bit rate. It differs from the previous standards in that it can support multiple block sizes from 4*4 to 32*32. A major part of all video or image transforms is Transform Coefficient Coding (TCC) [2]. HEVC uses the combination of Discrete Sine Transform (DST) and Discrete Cosine Transform (DCT) also called Hybrid Coding. DST is implied for intra prediction samples of size 4*4 and DCT for inter prediction and all the remaining block sizes of intra prediction that are supported [3]. Several architectures have been put forward to compute the Transform Coefficients (TCs). Some exploit pipelining, some make use of hardware sharing, others explore resources reuse. Each implementation style has its own pros and cons. We in this paper explore the scenario in which all these are selectively implemented according to the inputs requirements. Section II presents an overview of Transform Coefficient Coding (TCC) in High Efficiency Video Coding (HEVC). It also discusses a few of the existing architectures. Section III explains the proposed architecture in detail. Section IV elaborates and discusses the results obtained in different configurations. Finally Section V deals with the analysis, conclusions and comparison with the existing schemes. II. OVERVIEW OF TRANFORM CODING IN HEVC The HEVC transform uses both DCT and DST. The transform can be applied to block sizes from 8x8 to 32x32. HEVC has the additional benefit of supporting rectangular block sizes in which the size of row transform and column transform is different [4]. The integer transforms here are better approximations of Discrete Cosine Transform than H.264. The basis vectors are of equal energy so the need for compensation for different norms as in H.264 eliminates. For 4x4 intra prediction, HEVC uses DST especially for blocks coding with directional modes [5]. Several techniques have been put forward for transforming coefficient coding. The coding is meant to convert the coefficient into a form more suited for transmission. As the energy of the signal resides in the coefficient, the coding is a very crucial part. Any problem can lead to serious loss of energy as well as information. Transform coefficient coding in HEVC strives to achieve a balance between coding efficiency and practicality. As such, features and tools addressing the practical issues were adopted only if they improved coding efficiency or at worst, resulted in a slight degradation. The image is split into different blocks according to the previous coding standards. HEVC having larger block structures, has more flexible sub partitioning thus making it more adaptive to