ABSTRACT Photospheric bright points (BPs) fast movement is more effective than the slow movement that occurs most of the time in transmitting energy to the corona. The splitting and merging BPs tend to produce relatively large velocities. Therefore, our aim is to detect and track BP groups. After that, the attribute distribution of BP groups and the attribute evolution under possible high-speed movement are studied and analysed. We use a new BPs detection model (HBD-model) that can effectively detect weak BPs and layered BPs and then use similarity measures to judge the splitting and merging of BPs for tracking. The average accuracy of the BPs detection algorithm is 0.74, and the average accuracy of tracking is more than 85 per cent. For the BPs in the quiet Sun (QS) and active region (AR), a total of 62 388 evolutionary BPs were identified and tracked. The average velocities are more than 2 km s−1 in the QS and ARs. For the QS and ARs, the evolution process of splitting or merging BPs is basically the same. The results show that for the merged BPs, the area and velocity of the BPs are tiny in the initial stage, reach a peak in the middle stage, and then become smaller and disappear. The area, intensity contrast, and velocity of the BPs with splitting are relatively tiny at the beginning, become larger at the middle stage, reach the peak after splitting, and then become smaller and disappear rapidly.