9. Image geometric transformation algorithm for image processing and optics (bilinear interpolation)
9.0 Preface
Image geometric transformation is a very basic and practical skill in image processing, mainly including image displacement, scaling, mirroring, shearing, radial transformation, etc. In the process of spatially transforming images, the typical situation is to enlarge the image, When rotating, the image will be distorted. This is because in the image after transformation, there are some pixel positions that are not in the image before transformation. The method to deal with this problem is called image gray level interpolation.
There are many interpolation algorithms in images, commonly used are Nearest Neighbor, Bilinear, Bicubic, etc. Theoretically speaking, the effect of nearest neighbor interpolation is the worst, the effect of bicubic interpolation is the best, and the effect of bilinear interpolation is in between. However, for less stringent image interpolation, bilinear interpolation is usually sufficient.
9.0.1 Interpolation coordinate map
Interpolation, which finds the law based on the known data sequence (which can also be understood as a series of discrete points in the coordinates); and then according to the found law, numerically estimates the points that have no data records.
9.0.2 Interpolation Algorithm Application Scenario
The application is very extensive, and it is widely used in ISP-related data streams.