With the development of optical technology, higher requirements are put forward for the comprehensive performance of optical lenses. As an indispensable part of lens production, the core processing plays an important role in the comprehensive performance of lenses. The existing core processing requires workers to manually measure and adjust the core device, which requires high technical requirements for workers and slow processing efficiency. Based on Hough transform, this paper proposes a center bias detection algorithm suitable for centering processing to improve the measurement efficiency of centering processing. First, the image in the eyepiece of centering device is captured by an industrial camera, and the image is pre-processed by gratings, noise reduction, interference reduction and binarization according to image features. Then, according to Canny edge detection and Hough transform, the straight line of the cross image in the image is identified, and the intersection point is obtained according to the Angle of the straight line. The actual center bias of the lens is fitted by the least square method, and the center bias detection of the reflection measurement and the transmission measurement is realized. The experimental results show that the algorithm can detect the center bias under two working conditions, and the detection accuracy reaches sub-pixel level. The center bias detection algorithm used in this paper can improve the efficiency and accuracy of center machining.