Correlation detection can be used to extract weak signals such as those weakened by noise and attenuation. However, complex background noise contains both white and colored noise. The white noise is independent in the time domain, whereas the colored noise has a certain dependency, and the correlation detection algorithm cannot extract useful signals from the background mixed with colored noise. An improved correlation detection algorithm for whitening colored noise is discussed. Simulation experiments show that for laser signal amplitude similar to the noise amplitude, the correlation detection algorithm can extract the pulse signal, and the white noise background has a significantly improved signal-to-noise ratio of 2.3 times higher than the colored noise background. The experiment also proves that the whitening algorithm discussed can extract weak laser signals very well.