Hypertension causes cerebrovascular and cardiovascular diseases. To help prevent this, early detection requires continuous monitoring of blood pressure in daily life, which must be measured without contact. We previously proposed a noncontact technique for measuring blood pressure in which we evaluated the relationship between the variation in blood pressure caused by the cardiac-dominant pattern and facial skin temperature distribution using independent component analysis and multiple regression analysis. The mechanism of blood pressure change is roughly classified into cardiac-dominant patterns and vascular-dominant patterns. An elucidation of the relationship between these patterns and facial skin temperature distribution is the objective of this study. Two tests resulting in cardiac-dominant and vascular-dominant patterns were conducted. These common and different features related to each mechanism of BP change in each subject were then evaluated. The results of the tests show that similar features were extracted from the nasal region and contrasting feature quantities were extracted from the other facial regions. Hence, individual models were constructed for estimating blood pressure using the common features related to both mechanisms. A high coefficient of determination was obtained from these models, suggesting that models for estimating blood pressure caused by both mechanisms can be constructed using the proposed approach.