In this article, a novel necking detection and measurement method for automotive sheet metal components is proposed to detect and measure necking. The proposed method utilizes a point cloud registration-based approach to achieve better defect detection performance in the region of interest. In this method, an efficient registration algorithm called global feature-iterative closest point is introduced, which exhibits excellent performance for complex surfaces, such as sheet metal parts. Subsequently, an algorithm called normal vector propagation is proposed for defect point cloud detection and extraction, enabling the acquisition of comprehensive necking information. Finally, a necking dimension measurement method, referred to as triangle mesh–registration distance elimination, is introduced to measure the surface area and depth of necking. Experimental investigations were conducted on two sheet metal components with necking, and comparisons were made with other methods. The results demonstrate the practicality and effectiveness of this proposed method.