To improve the efficiency and performance of signal recognitions, based on the analysis of the short-time signal's ambiguity function in the time-lag domain, the kernel function of Choi-Williams Distribution(CWD) was improved for short-time signals. After several short-time sections of the signal to be recognized, a new recognition approach was realized by comparing them in the time frequency domain using the improved CWD.The experimental results showed that the approach not only reduces the calculation and improves the agility of the recognition but also maintains the high performance of CWD under noise environments.
