Morphological operators leverage various functions to address tasks such as noise reduction, contour detection, structure identification, and shape regularization, offering simplicity and effectiveness. Particularly, morphological filters are commonly employed in nonlinear operations, especially for tasks like acquisition, binarization, and filter generation. Our approach introduces a novel iterative method for performing morphological filtering on binary images, aiming for superior results. This method involves defining counterparts to basic morphological operators, erosion, and dilation, to analyse filter operations. While our proposed technique enhances outcomes, it's worth noting that traditional image processing methods often struggle with real-time image calculations. To overcome this, we utilize M4K blocks instead of conventional registers to accommodate pixels that may exceed register capacity. Additionally, shift registers are employed to manage address data efficiently in a pipeline, thereby speeding up processing. We have implemented opening, dilation, edge, erosion, and closing detection with Verilog HDL. Users have the flexibility to empirically select thresholds, including prolonged exposure and Sobel sensitivity, to eliminate noise through external inputs.
Gupta et al. (Sun,) studied this question.