General-purpose processors are versatile tools capable of performing a wide range of tasks, but they lack the efficiency needed for computationally intensive applications. High-resolution image processing and deep learning-based computer vision models, which demand substantial computational power, have driven the adoption of specialized hardware accelerators such as DSPs, GPUs, TPUs, VPUs, NPUs, and FPGAs. Among these, Xilinx’s all-programmable SoCs, combining ARM processors with mid-scale FPGAs, alongside the Vitis platform and Vision Library, provide robust solutions for high-performance image processing. This study harnesses these advanced technologies to accelerate morphological image processing on the PYNQ-Z1 and Kria KV260 platforms. Using a hardware-software co-design methodology with Vitis HLS and Vivado, we implemented essential morphological operations—dilation, erosion, opening, closing, and morphological gradient—on 4K-resolution images. Experimental results demonstrate substantial performance improvements, achieving over 4 \(\times \) on the Kria KV260 for computationally complex tasks, all while keeping power consumption below 3.1 W. The project resources made publicly accessible on GitHub ( https://github.com/SEHAIRIKamal/Xilinx-PYNQ-Image-Processing ).