Blood pressure (BP) measurement data is an important indication of health and quality of life in clinical medicine and daily life. However, conventional measurement does not provide continuous monitoring data and sometimes is considered as inconvenient. With the ever increased urge for cuff-less continuous BP measurement, novel methods based on pulse transit time (PTT) obtained from the photoplethysmogram (PPG) and electrocardiogram (ECG) signals have gained its popularity. However, the collection of ECG signals involves the application of electrodes and inconvenience due to lengthy continuous measurement. In contrast, the collection of PPG signals is comparatively simpler and easier, therefore, novel methods that extract features from PPG signal are receiving more attentions. However, previous studies only focus on the features extracted from the PPG signal and did not include the physiological characteristics, which can serve as important predictors for BP. To improve the accuracy in the estimation of BP based on the PPG signal, this study not only extracts features from the PPG signal, but also includes the physiological characteristics of total 191 sets of subjects, such as height, weight, and age. After pre-processing the raw PPG signal, different machine learning methods are used to estimate the diastolic blood pressure (DBP) and the systolic blood pressure (SBP). The mean absolute error of DBP and SBP are 4.13 mmHg and 9.18 mmHg respectively. The results complied with the British Hypertension Society (BHS) standards and the implementation of physiological characteristics improved the accuracy of BP estimation.