The normalized transformation consists of two parts, the translation of the image coordinates and the scaling of the scales. The transformation of normalization can not only improve the accuracy of processing results, but also eliminate the influence of the selection of image scale and coordinate origin on the final result of the algorithm by selecting a standard coordinate system beforehand.
Steps to normalization a transformation:
- Pan the points so that the hearts (centroid) of these points move to the origin
- Scale scales so that the average distance from the points to the origin is
The normalization of data is a must in some algorithms, especially for some less well-established problems, such as: The calculation of basic matrices and the DLT algorithm of Sanjiao tensor.
Solution of normalized transformation matrix T:
Yes, average.
The scale S is
Normalized Transformation normalizing transformations