Abstract: The automotive industry has a large number of multi-hole parts with different shapes. A template matching algorithm based on feature triangle has been proposed for template matching of multi-hole parts. The matching holes are selected by the area feature and the distance feature. The constrained least-squares fitting circles is used to calculate the precise center of the matching holes to construct the feature triangle. According to the rotation invariance and translation invariance of the triangle, use the feature triangle to extract the matching center and the matching angle for affine transformation to complete the template matching and registration of the part. Experimental results show that the template matching algorithm is suitable for parts with different hole numbers. The algorithm has high tolerance of the deformation of the parts, and is not easily affected by the defects such as partial missing and blocking of matching holes. The algorithm has a fast matching speed. It can meet the requirements of industrial parts defect detection.