At present, water shortage in agriculture becomes more and more serious. This situation makes it necessary to develop precision irrigation, which needs to obtain the accurate crop water stress in advance. However, this signal is too weak to be detected easily. Wavelet analysis has been widely used in the signal processing area for almost two decades due to its excellent time-frequency analysis ability. Therefore, the wavelet decomposition and reconstruction technique is applied to reduce the noises of experimental data collected from corn plants in a farmland. Finally, data analysis results show that wavelet denoising is effective to achieve the weak signal extraction.

At present, the agriculture in China is severely short of water, which turns water-saving agriculture into a must due. The basic principal of water-saving agriculture is to monitor the crop water potential on time and establish a closed-loop control system. The reality of accurate irrigating is due to the water requirement of the crop. In this way, obtaining the accurate water potential is the key technology for water-saving agriculture. However, the crop water potential and sophisticated field surroundings are relatively weak signals. In this paper, the signals are pretreated. Then the authors identify the signals of the crop water potential from the noised-signals and predict the grow trends by wavelet decomposition and reconstruction. The simulation result proves that the method could basically reflect the trend of desired signals, which is an effective analysis.