05 Geometric Transformation.pptVIP

5 Geometric Transformation By Wang Qiang College of Computer Science Zhejiang University Outline Spatial Transformation General Spatial Transformation Some Basic Geometric Transformations Translation Scale Rotation Gray-level Interpolation Implementation of Spatial Transformation Summary 5.1 Spatial Transformation A geometric transformation for image restoration consists two basic operations: A spatial transformation Gray-level interpolation 5.1.1 General Spatial Transformation Suppose an image f with pixel coordinates (x, y) undergoes geometric distortion to produce an image g with coordinates (x, y). The transformation may be expressed as Tiepoints: a subset of pixels whose location in the input (distorted) and output (corrected) images is known precisely. The transformations are usually approximated using polynomials: To correct the image more precisely, quadratics are often used. 5.1.2 Some basic spatial transformation Expressed in homogeneous coordinates and matrix Composite transformation: expressed by multiplication of matrices, such as the rotation around a point (x0, y0): 5.11.2 Gray-Level Interpolation x and y may not be integers. Zero-order interpolation 5.3 Implementation of Spatial Transformation Forward mapping: Backward mapping: Summary Geometric Transformations Two steps * College of Computer Science, Zhejiang University * If r and s are known analytically, recovering f from distorted image g by applying the transformation in reverse might be possible theoretically. However, … The tiny distortion are usually approximated using linear transformations. The coefficients of the supposed transformation formula is solved using the coordinates of tiepoints. 1. Identical transformation 2. Translation 3. Rotation 4. Scale 1. Translation 2. Rotation 3. Scale A five degree counter-clockwise rotation with bilinear interpolation (x0, y0) is nearest integer coordinate neighbor to (x, y). bilinear interpolation fall among grid points Disadvantages: May map to