Synthesis of Area - E cient andHigh - Throughput Rate

We propose two design methodologies for synthesis of area-eecient Data Format Converters (DFCs) with high throughput rate. DFCs are grouped into various classes according to the speciication of design parameters. The rst design methodology is suitable for design of many representative classes of DFCs. The designs using this methodology are based on a two-dimensional architecture. They have maximum throughput rate and are area-eecient. Various design examples are shown to demonstrate improved performance, exibility and usefulness of this design methodology. For several representative problems, the area requirements of our designs are compared against those obtained by earlier design methodologies. For all the problems considered, this methodology leads to compact designs. The second design methodology employs an architecture using dual buuers. The simple and regular architecture using dual buuers leads to area-eecient DFCs. The design procedure using this methodology is simple and can reduce the design eeort in many applications. Data format converter (DFC) is a hardware interface module employed to reorganize the format of the transferred data between various processing modules with diierent I/O format requirements 1, 2, 3]. Data format conversion is required in many signal/image processing applications such as two-dimensional Discrete Fourier Transform 4], low-level vision applications, and video coding/decoding (codec) 5, 6], video pre-/post-processing, image scaling etc.. For example , consider the design of a VLSI system for an 1 application based on a video compression standard such as H.263 7, 8, 9]. The system consists of two parts, the video processing part and the video compression part. In the video processing part, the format of the image grabbed from a video camera and the format of the image required in the video compression system are different. For pre-processing and post-processing of video, data format conversion is required during image scaling. Format conversion to and from the standard image formats such as CIF and QCIF is essentially a data format conversion problem. In the video compression part, there are various processing modules, and data format conversion is required within the modules as well as between successive modules. Specii-cally, data format conversion is required during the computation of two-dimensional Discrete Cosine Transform, and between the Quantizer and the Variable Length Coding module. Data format conversion is also required to perform diier-ent modes of the motion estimation module such as half-pixel computation and multi-resolution search (when a fast search algorithm is used) 7, 8]. The design of data format converters …