In track and field sports, athletes' movements are highly personalized and random, which makes it difficult to directly compare and analyze the similarities between movements made by different athletes or the same athlete in different situations. Aiming at the problem of similarity analysis of track and field athletes' movements, this study designs and implements an analysis system based on the Dynamic Time Warping (DTW) algorithm. The overall architecture of the system includes four modules: data preprocessing, time series alignment, distance calculation, and result output and analysis. The data preprocessing module extracts key features and converts them into time series format. The time series alignment module uses the DTW algorithm to achieve accurate alignment of time series. The distance calculation module quantitatively evaluates the similarity of actions by calculating the Euclidean distance between the aligned time series. The result output and analysis module helps users understand the differences and similarities between athletes' movements. Compared to traditional template matching methods, the accuracy of motion sequence alignment and computational efficiency have been significantly improved through the optimization of the DTW-based system. Experimental results show that the system reduces alignment error, enhances similarity measurement accuracy (with the correlation coefficient increasing from 0.68 to 0.76), and significantly optimizes computational performance while reducing processing time and memory consumption. Additionally, the system enables dynamic tracking of athletes' movement variations across different scenarios, providing real-time technical optimization suggestions and offering better support for personalized training strategies.