Multidimensional digital signal processing systems for the next generation visual communication

本研究针对下一代视觉通信中的多维自适应滤波器进行了研究，在LMS算法的基础上提出了格型和直接型的二维自适应滤波结构和二维自适应算法。将这些滤波器结构应用于系统辨识和自适应谱线增强器中，以显示它们的快速收敛速度和较高的降噪能力。基于LMS算法，提出了并行可分离的二维自适应滤波器结构。这些自适应滤波器最重要的特点是易于进行稳定性测试和可并行实现。这些自适应滤波器的计算量很少，例如O(N)-O(n^2)，而传统的自适应滤波器需要O(n^2)-O(n^3)，其中n是自适应滤波器的阶数。

This research studies multidimensional adaptive filters for the next generation visual communication.2-D adaptive filter structures with lattice and direct forms and 2-D adaptive algorithms are proposed based on the LMS algorithm. These filter structures are applied to system identification and adaptive line enhancer in order to demonstrate their fast convergence speed and high ability of noise reduction.Parallel separable 2-D adaptive filter structures are proposed based on the LMS algorithm. The most important features of these adaptive filters are ease of filter stability test and the possibility of arallel implementation. These adaptive filters need very few computational requirements such as O (n) - O (n^2) computations while conventional adaptive filters need O (n^2) - O (n^3) , where n is the order of adaptive filters.