Defect detection of solar cells in electroluminescence images using Fourier image reconstruction

Abstract

Solar power is an attractive alternative source of electricity. Solar cells, which form the basis of a solar power system, are mainly based on crystalline silicon. Many defects cannot be visually observed with the conventional CCD imaging system. This paper presents defect inspection of multicrystalline solar cells in electroluminescence (EL) images. A solar cell charged with electrical current emits infrared light, whose intensity is lower at intrinsic crystal grain boundaries and extrinsic defects of small cracks, breaks, and finger interruptions. The EL image can distinctly highlight barely visible defects as dark objects, but it also shows random dark regions in the background, which makes automatic inspection in EL images very difficult.A self-reference scheme based on the Fourier image reconstruction technique is proposed for defect detection of solar cells with EL images. The target defects appear as line- or bar-shaped objects in the EL image. The Fourier image reconstruction process is applied to remove the possible defects by setting the frequency components associated with the line- and bar-shaped defects to zero and then back-transforming the spectral image into a spatial image. The defect region can then be easily identified by evaluating the gray-level differences between the original image and its reconstructed image. The reference image is generated from the inspection image itself and, thus, can accommodate random inhomogeneous backgrounds. Experimental results on a set of various solar cells have shown that the proposed method performs effectively for detecting small cracks, breaks, and finger interruptions. The computation time of the proposed method is also fast, making it suitable for practical implementation. It takes only 0.29s to inspect a whole solar cell image with a size of 550×550pixels.