To address the problems of low accuracy and slow speed of manual inspection in the production process of lithium battery, a defect detection method for lithium battery pole piece is proposed based on Nlm non-local mean filtering and improved Grabcut image segmentation algorithm. For the defect images, firstly, remove the useless parts and obtain the pole piece region containing only the defect parts; secondly, Nlm non-local mean filtering is applied for image denoising, and the image is enhanced by using the single-scale retinal algorithm SSR, which obtains an image with low noise and high contrast between defects and background; Finally, the defects are extracted using the improved Grabcut image segmentation algorithm, and the defects are Finally, the defects are extracted using the improved Grabcut image segmentation algorithm, and the morphological corrosion and expansion operations are performed to optimize the image and complete the calibration of the defect contour. The experimental results show that the algorithm is able to accurately detect defects such as metal leakage, white spots, black spots, decarburisation and striations, which are difficult to detect against the background of the coating area of the pole piece.