A pixel matching method in on-line three-dimensional measurement based on normalized equiphase plane was proposed. Only one fixed sinusoidal fringe was needed to project on a measured object moving on the pipeline. While the object was coming to some positions with a certain adjacent displacement, the deformed patterns modulated by the object were captured synchronously by the CCD camera. Dealing the captured deformed patterns with FTP method, the phase information of the object in those positions could be predicted and normalized, and the normalized equiphase plane was transformed into integer as the mask to assist the pixel matching, which not only realized the one-to-one corresponding of the object, but normalization reduced the error introduced by these differences of the object′s unwrapped phase in different positions caused by the object′s movement, further binarization of the phase feature improved the pixel matching′s speed as well. The simulation result of the "peaks" function object with the max height 8 mm showed the RMS was 0.021 mm. Meanwhile, as for time spent of the pixel matching, the average time spent with the proposed method was about 2 multiples faster than the method based on phase prediction. The object measurement also showed its feasibility and validity of the proposed method. The proposed method could not only guarantee the accuracy of the online 3D measurement, but improve the measurement speed apparently.