The invention provides a polysilicon solar battery piece appearance contamination defect detection method. The method mainly comprises two steps, wherein the first step comprises pretreatment of a battery piece and acquisition of an interested region, and the subsequent operations are performed by taking the first step as the foundation; and the second step comprises recognizing and marking of thecontamination regions. Pretreatment comprises the steps of dividing collected three-channel images into R channel images, G channel images and B channel images, and selecting the R channel images with relatively high defect and background comparison degree for performing subsequent treatment; performing smoothening treatment on the selected R channel images, suppressing interference from image noise and high frequency components, next, performing rotary correction operation on the images, enabling the image edges to be parallel to a coordinate system, next, performing fixed threshold value image cutting, setting upper and lower critical values according to the area, and removing interference from a conveyor belt region on the image processing; and next, performing morphology closed-loop operation on the images, removing small sized black holes generated by insufficient threshold value cutting, and finally realizing polysilicon solar battery piece appearance contamination defect detection.