The system for collecting plant micro-environment and physiological parameters was featured with large amount of data, storage and transmission difficulties, which restricted the long-term monitoring of ecological information. A new method for collecting plant micro-environment and physiological parameters based on compressive sensing was proposed. The bases of discrete Fourier transformation (DFT) and discrete cosine transformation (DCT) were selected as sparse bases under which the working performance of system was tested. Through the experimental analysis, DCT had smaller sparse error and DFT had better denoising effect, lower observed dimension, higher compression ratio under condition of the same sparsity. With the increase of sparsity, both sparse error and compression ratio were decreased. In the meantime, both observed dimensions were increased. In addition, data compression ratio and power consumption of the DFT compressive sensing system were tested based on ARM. The conclusion was that the data compression ratio was 4.24, and 13.62% power consumption was saved compared with traditional collecting system for plant micro-environment and physiological parameters. In conclusion, the method proposed had advantage of compressing data, saving storage, decreasing data transmission, which can extend working hours of the system by decreasing power consumption and make an important effect on the development of internet of things in agriculture and forestry.