The invention discloses an online rapid chicken ash content detection method. Hyper-spectral images of a chicken sample are collected and corrected, the obtained spectral images are preprocessed, thetarget regions are identified, the average spectral data of spectral images is extracted, and the extracted spectral data is substituted into a formula. 13 optimal wavelengths are extracted from 486 full wave bands; a large amount of redundant information is eliminated, useful information is extracted, the calculated amount of data analysis is reduced, thus the precision of a partial least squaresmodel is improved, and the requirements of large scale online meat production can be satisfied. Compared with the prior art, the method has the advantages that a tested sample does not need to be preprocessed, the sample is not contacted, is only spectrally scanned, and is not destroyed; no chemical reagent is used, the method is green, cost is saved, the operation is easy, time is saved, and large scale online detection of chicken ash content is realized.