In order to aviod the complexity of extracting feature points from 2D images and performing 3D reconstruction, and to meet the requirement of high precision of 3D measurement index of chip pins, this paper proposes a horizontal calibration method of point clouds based on difference of normal vectors by using point cloud processing technology, and combines this method with the minimum outer bounding box algorithm to eliminate the point clouds of chip internal pins. After clustering and segmenting the remaining point clouds, the plane equation parameters of each external pin, upper and lower surfaces of the chip are fitted by using the sampling consistency algorithm, and the midpoint of each pin area is solved. Calculate packege height, stand off and foot angle through the data obtained above. Finally, the effectiveness of this chip measurement scheme is verified by experiments.