The Applications and Properties of the TDS Algorithm

The 2-D smoothing (TDS) algorithm is a powerful tool for smoothing and filtering 2-D sequences, serving a crucial role in tasks such as image processing and filtering. Despite its significance, the theoretical properties of the TDS algorithm have not been thoroughly explored in the current literature. In this article, we present a comprehensive analysis of the TDS algorithm, elucidating its mathematical properties and proposing innovative models for its application in image processing. First, we provide an equivalent description of the TDS algorithm and demonstrate that the trend sequence makes the loss function reach the global minimum. Regarding the convergence of the TDS algorithm, we demonstrate the convergence of both trend and fluctuation sequences. Specifically, as the global smoothing parameter tends to infinity, both sequences converge to a deterministic sequence that is independent of the global smoothing parameter. In this case, the TDS algorithm becomes equivalent to computing the trend sequence via the least squares method (LSM). Subsequently, we reveal the smoothing mechanism of the TDS algorithm, which attenuates the energy of the original sequence in the transform domain. Furthermore, we show that the forward transform kernel of the TDS algorithm is a separable orthogonal transform. In addition, we explore the intrinsic relationship between the trend and fluctuation sequences. Notably, an insightful result is that the fluctuation sequence is the trend sequence of the sequence obtained by applying the characteristic lag operator polynomial to the original sequence. Building on these insights, we propose several application scenarios and models for the TDS algorithm in image processing, such as image smoothing, high-frequency extraction, edge detection, and enhancement. To validate the effectiveness of the TDS algorithm, we present numerical simulations and image processing experiments, which demonstrate the correctness of the proposed theoretical framework and the superior performance of TDS in 2-D filtering tasks. This work lays a solid theoretical foundation for the practical application of the TDS algorithm, offering novel methodologies and insights for image processing, 2-D filtering, wireless communication, and computer vision.