In this short paper, we introduce a parameterized shape model, rose curve model. An analytical solution for estimating rose curve parameters from a binary silhouette or a probability map is derived. This analytical method finds the global optimum directly and therefore is fast and reliable. Two similarity invariant shape features, which measures the concavity and circular frequency of the shape can be derived from the six parameters of the rose curve. We apply the rose curve model to approximately segmenting flower images, primarily for testing the analytic parameter estimation method. Experiments on a database of 180 flower images from 30 species show that the rose curve is an excellent shape model for many flower species and the analytical parameter estimation method locates the flower regions well.