Aiming at the problem that the online detection of chromatic aberration of small box cigarette case on the quality inspection machine is slow, an online detection method of cigarette label chromatic aberration based on k-means clustering is proposed. Based on image processing and machine vision, this method identifies the standard sample collected by the CCD camera and the main background color of the cigarette scale to be detected, and then converts through the color space, and finally performs the detection of chromatic aberration, and the chromatic aberration is less than 4.0 for the qualified cigarette case. Experimental results show that because there is no need for background removal and image registration, the detection speed of this method is better than that of other algorithms, and the accuracy can meet the actual production needs.