Advanced topographic laser altimeter system (ATLAS) can provide scientific data for global canopy height measurement. However, due to the characteristics of background noise in the photon data, the traditional algorithm does not study for the forest coverage area under weak beam conditions and there was still few photon cloud noise filtering algorithm can evaluate the accuracy of noise filtering under the condition of weak beam in forest research area. In order to quantify the accuracy of the photon cloud noise filtering algorithm in the forest research area under the condition of weak beam, the accuracies of the local distance statistical algorithm, the densitybased spatial clustering of applications with noise (DBSCAN) algorithm and the particle swarm optimization (PSO)-DBSCAN algorithm for photon cloud noise filtering experiments in forest areas under weak beam conditions were studied, and the influence of different characteristics on noise filtering results was analyzed. The results were as follows: the results showed that PSO-DBSCAN algorithm had the accuracy of noise filtering as 0.95 in the forest area under weak beam conditions, which met the accuracy of photon cloud noise filtering requirements, and the algorithm performed better than the local distance statistical algorithm and the DBSCAN algorithm. The solar elevation angle had greater impact on the noise filtering algorithm than the terrain slope and vegetation coverage.