Traditional cuff-based sphygmomanometers can cause discomfort during blood pressure measurements, rendering them unsuitable for patients with skin conditions as well as for real-time monitoring during sleep. To address this issue, this study proposes a cuff-less blood pressure estimation model that leverages the morphological features of photoplethysmogram (PPG) waveforms. By utilizing the dynamic variations in short-term pulse waveforms, the model estimates blood pressure, thereby replacing conventional measurement methods. Using data from 500 subjects in the Cuff-Less Blood Pressure Estimation Dataset provided by the University of California, Irvine Machine Learning Repository, a deep learning model based on the Residual Network (ResNet) architecture was developed to effectively extract temporal features from PPG signals and perform regression predictions for systolic blood pressure (SBP) and diastolic blood pressure (DBP). Experimental results reveal that the proposed model achieved a mean absolute error (MAE) of 2.317 mmHg for SBP and 1.230 mmHg for DBP on the test set, with corresponding accuracies of 98.35% and 98.07%, respectively, thereby meeting international medical standards for blood pressure measurement. These findings support the feasibility of using PPG signals for cuff-less and calibration-free blood pressure estimation, highlighting significant potential for improving the accuracy and clinical applicability of non-invasive blood pressure monitoring technologies, especially when integrated with cardiovascular dynamics information in the future.