The four-chamber view of heart plays an important role in clinical diagnosis of cardiac disease. According to the geometric structure of heart, we propose a method to automatic acquisition of the four-chamber view in volume data. First, the contrast-medium enhanced part of dual source CT data is segmented and morphological operations are applied. Second, the Dijkstra Algorithm considering 3D Euler Distance as weighting factor is introduced to extract the center line of atrium and ventricle. Finally, the four-chamber view of heart is estimated based on the 3D Euclidean Distance and the geometric characteristics of the center line. In this paper, 40 patients (age from 49 to 81) are examined, and the recognition rate is 91.25%.