The number of scanner stations used to acquire point cloud data is limited, resulting in poor data registration. As a result, a cloud point block registration approach was proposed that took into account the distance between the point and the surface. When registering point cloud data, the invariant angle, length, and area of the two groups of point cloud data were affine transformed, and then the block registration parameters of point cloud data were determined. A finite hybrid model of point cloud data was created based on the coplane four-point nonuniqueness during the affine translation. On this basis, the point cloud data block registration algorithm was designed. Experimental results prove that the proposed method has great advantages in texture alignment, registration accuracy and registration time, so it is able to effectively improve the registration effect of point cloud data. The point cloud data block registration algorithm was built on this foundation. Experiments show that the suggested method has significant improvements in texture alignment, registration accuracy, and registration time, indicating that it can significantly improve point cloud data registration.