Soft ROI Created Through Sigmoid Functions
摘要
This chapter proposes to create a new type of soft ROI (window) for signal and image processing through sigmoid functions. The S-shaped ROI boundary enables the sidelobes of a hard rectangular ROI to be largely reduced, while the relatively flat ROI area makes it better to maintain the raw discrete signal resolution compared to the common bell-shaped ROIs (windows). At the same time, such a new type of ROI is infinitely differentiable and parameterizable, making it powerful and adaptable to meet many types of application requirements.