The method of counting for circular ring objects in frequency domain is proposed. A binarized image with circular rings is processed by Fourier transformation and filtered in frequency domain, then inverse Fourier transformation is carried on. After selecting a suitable threshold, the circular rings become solid particles individually. The number of circular rings is obtained through counting the solid particles. The mathematical model is established and the experimental results for both simulated and real images are also given. The method is robust to noise and the detection for partially occluded circular rings is also available. The method can also be applied to the situation that some objects are overlapped. The method is more effective when the size of the objects in the image is similar to each other.