The automation and accuracy of collecting phenotypic parameters had been a difficulty in traditional flower plants. In this paper, we proposed a method for the segmentation and phenotypic analysis of anthurium organs based on three-dimensional point clouds. Firstly, the three-dimensional point cloud model of anthurium was obtained based on the structure from motion algorithm with application of the background removal and point cloud denoising. Then, an automatic segmentation algorithm for anthurium organs based on color information was proposed to automatically extract individual anthurium organs point cloud. Combining the two attributes of RGB and curvature of the point cloud, an anthurium organ classification rule was proposed to achieve automatic classification of anthurium organs. On this basis, an effective method for calculating the phenotypic parameters of anthurium was given to accurately measure four key traits, namely plant height, crown width, bract diameter and number of spathes. The results showed that this method could achieve automatic analysis of the key phenotypes of anthurium. The root mean square errors of the measurement results of plant height, crown width, bract diameter and number of spathes obtained by this method compared with the manual measurement results were 2.115, 1.428, 0.214 cm and 0.272, respectively, and the average absolute percentage errors were 4.445%, 3.765%, 2.295% and 3.783%, respectively. The overall error was lower than that of other methods. It can be seen that this method could be used to extract phenotypic parameters of anthurium flowers with complex structures and irregular morphology in practice.