The invention relates to an irregular workpiece detection method based on machine vision, and the method comprises the steps: obtaining an irregular workpiece image, carrying out the edge detection, converting the edge of the workpiece image into a polar coordinate system, employing the intersection of the polar coordinate system in the horizontal direction of different theta connecting lines as a new two-dimensional plane through Hough transformation, according to the information gain of all connecting line intersection points in the two-dimensional plane, judging the intersection points to obtain missing detection points, obtaining a missing detection area according to the missing detection points, performing bubble detection on the workpiece, if the missing detection area does not have bubbles, determining that the workpiece does not have defects, and if the missing detection area has bubbles, obtaining an edge area of the bubbles in the missing detection area; and according to the volume and position changes of all bubbles in the adjacent frames and the single-frame image are analyzed according to the marginal areas of the bubbles, determining the accurate position of the missing detection defect, so the defect position of the irregular detection workpiece is more accurately positioned based on machine vision and comprehensive analysis, the detection precision is high, and the practicability is high.