The invention relates to an EL (Electro Luminescence) image detection and defect identification method for solar cells. The EL image detection and defect identification method for solar cells includesthe following steps: (1) obtaining a to-be-detected EL image of a solar cell, positioning a gate line and performing region division; (2) deleting the gate line region, recombining the image, calculating the image gray value and performing two-dimensional construction (3) calculating the inter-class dispersion matrix of a particle swarm to determine the current optimal position; (4) updating theoptimal individual of the particle swarm and the historical optimal individual of the particles; (5) generating a new chaotic variable by means of a chaotic model; (6) updating the position and velocity of all particles of the particle swarm, recalculating until the number of iterations is reached; and (7) obtaining a defect image of the cell through segmenting according to the obtained optimal position, and performing defect identification. The EL image detection and defect identification method for solar cells is simple to implement, is high in operation velocity, can be adapted to differenttypes of defects, and can prevent local convergence by segmenting the EL image of the cell by mean of the chaotic particle swarm, thus obtaining a more accurate defect image.