Most of existing algorithms for detecting tropical cyclones (TCs) in global climate model employed multiple detectors, and screened the field grid by grid by thresholding, while a quantitative assessment of the performance of these algorithms has been infrequent. A new method based on mathematical morphology for detecting and tracking TC in gridded reanalysis data is proposed and evaluated. The method begins by transforming the field of lower-tropospheric (850 hPa) relative vorticity from the reanalysis data into a binary image using a defined critical vorticity threshold. The images are then processed by connected component labeling and erosion operations until inner cores emerge as the TC seeds. Simple rules for connecting the seeds to form TC tracks are proposed. Twenty years of ERA5 data and the China Meteorological Administration best track data (CMA) in 2002–2021 are used to develop and evaluate the method for the Western North Pacific basin (WNP). The detected TC events are examined track-by-track with the observed ones. The performance of the method is further evaluated in terms of the statistics of TC occurrence, track density, landfall location, and translational speed. It is demonstrated that the proposed method can reproduce the TC tracks for both over ocean and after landfall with good accuracy and efficiency. The source of erroneous tracking and limitation of the proposed method are discussed. Given its satisfactory performance in tracking TCs in the WNP basin, it is promising to be applied to the study of TC changes in a warming climate.