Particle detection of fluorescent images has become an indispensable tool in biological research. Here a simple and fast method for biological particles detection with high efficiency and accuracy, improved wavelet transform (IWT) was introduced. IWT originates from wavelet multiscale products (WMP). However, it resolves the problems in WMP and is more adaptive in dealing with different types of images. The performance of IWT, WMP and MSVST (multiscale variance stabilizing transform) was quantitatively evaluate by using both synthetic and real fluorescence images. Experimental results show that IWT performs much better than WMP in most cases, and has comparable results with the much complicated algorithm, MSVST. Besides, IWT is 20% faster than MSVST when processing the same images. Therefore, it was concluded that IWT can be generally used for the automatic detection of different kinds of biological particles, and the simplicity and accuracy make it a better choice for fluorescent image analysis.