Accurate wound segmentation is a pivotal task in medical image analysis, aiding clinicians in efficient diagnosis and treatment outcome monitoring. In this paper, we introduce an innovative approach for fully automatic wound segmentation utilizing a hybrid architecture combining two U-Net architectures. The proposed model takes advantage on the strengths of original U-Net architecture to achieve enhanced accuracy in segmenting wound regions. Our proposed method achieved 97% accuracy, 0.97 precision, 0.93 recall, 0.90 IoU score, and 0.95 dice coefficient similarity score. We built an annotated wound image dataset consisting of 365 images from 110 patients to train and test our model. We demonstrated the effectiveness and mobility of our method by conducting comprehensive experiments and analyses on various segmentation neural networks. Tested on diversified wound types, our model offers robustness and adaptability for varying wound appearances. The result promises to strengthen clinical decision-making and treatment monitoring through improved wound assessment.