Down-Conversion for Improved Picture Quality
Down-conversion is a fundamental part of the digital television technology that provides compatibility of the emerging HDTV equipment with the traditional standard television broadcasting, and it can also be applied to a variety of other consumer electronics products features or requires variable resolution video image feeds. In this project, we design down/up conversion filters using frequency synthesis and spatial domain methods, and compare the performances of various conversion techniques in terms of image quality. We also investigate the use of extracted side information to improve picture quality and effects on the system architecture.
Background & Objective: The project is initiated to conduct research on new algorithms for memory reduction within a video decoder. Not only are memory requirements reduced, but implementation costs are also reduced due to impact on memory bandwidth and clock rate. Two key technologies have been the major focus: down-conversion and motion compensation. These new algorithms are basic components of a low-cost video decoder and may be used in a wide range of image/video products.
Technical Discussion: In order to reduce the size an image without causing aliasing errors, the image is filtered initially to limit the bandwidth of the input signal spectrum. A low-pass filter tuned at the Nyquist frequency, which is the half of the signal bandwidth, removes the high frequency coefficients. Then, the filtered image then can be sub-sampled accordingly. The filters used for down-conversion are based on the concept of frequency synthesis, and the filters used to perform the up-conversion are determined by an optimal least-squares solution. The combination of these techniques allows us to achieve significant reductions in the amount of observable drift compared to previously published methods.
Outside Collaborations: Lucent and Cadence
Contacts:
Anthony Vetro
Fatih Porikli
Technology Areas:
Advanced Digital Television
Digital Communications
Modification Date: September 12, 2007
