Support tucker machines based bubble defect detection of lithium-ion polymer cell sheets

Recently Lithium-ion polymer (LiPo) battery attracts more interests in both technology and application for its high energy density and easy to be manufactured in different shapes. The LiPo battery quality is essential for all the applications, and the defect detection of LiPo cell sheets in the automatic production line is critical to battery quality control. After capturing the images of two sides of LiPo cell sheets with controlled manipulators in the automatic line, support Tucker machines (STM) method based on the tensor is applied to detect bubble defects in Lithium-ion polymer cell sheets. The preprocessing of the cell sheet images and the proposed STM based method are detailed. The experimental results demonstrate that the proposed STM based bubble defect detection method is more efficient than other common learning based methods, and the proposed STM based method is potential for classification in image applications.