We present a measurement method for obtaining three-dimensional (3-D) images of the thoracic and abdominal body surface during respiration. This method generates a color pattern composed of R, G, and B cosine stripe patterns to obtain a single image-based 3-D measurement. The 3-D wavelet transform is adopted to extract wrapped phases from a stripe image sequence to improve the accuracy of the wrapped phase extraction. Then, a two-frequency temporal phase unwrapping formula is proposed and extended to a multifrequency formula according to the geometric relationship between the wrapped and the unwrapped phase curves, which enhances its anti-interference performance and calculation speed. Simulations and body measurements reveal that the proposed method can effectively measure the 3-D body surface during respiration. By using multiple characteristic parameters, this method overcomes the limitation of only using feature points to represent respiratory movement on the thoracic and abdominal body surface. This method has important applications for tracking respiratory motion during clinical radiotherapy.