Documentation › Documentation Library › Tilt / Shear Correction
Edit & Preprocess
Tilt / Shear Correction
Tilt / Shear Correction corrects horizontal geometric skew in a hyperspectral image.
Overview
Tilt / Shear Correction corrects horizontal geometric skew in a hyperspectral image.
The operation applies a horizontal shear transformation to the spatial dimensions of the hyperspectral cube.
The same geometric transformation is applied to every spectral band so that spatial registration between wavelengths is maintained.
Tilt / Shear Correction modifies the current working dataset. The original dataset remains available through the IDCubePro Reset function.
WHAT IS A SHEAR TRANSFORMATION?
A shear changes image geometry by shifting image rows horizontally relative to one another.
As the amount of shear increases, vertical structures become progressively more slanted in one direction or the other.
Applying the opposite shear can therefore straighten structures that appear geometrically skewed.
Tilt / Shear Is Different From Rotation
Rotation turns the entire image around a point.
Shear changes the relative horizontal position of image rows.
If the entire image is simply oriented at the wrong angle, use Rotate by Angle instead.
If structures are slanted while the overall image orientation is otherwise correct, Tilt / Shear Correction may be more appropriate.
When To Use Tilt / Shear Correction
Use this tool when an image contains systematic horizontal geometric skew.
A sample that appears progressively displaced from top to bottom.
Geometric skew introduced by an imaging or scanning configuration.
A rectangular structure that appears slanted even though the overall image should not be rotated.
Image geometry that requires shear correction before subsequent spatial analysis.
Do not use shear simply as a cosmetic replacement for rotation.
Tilt / Shear Correction can be opened from the image orientation controls.
It can also be launched from Favorites when Tilt / Shear Correction has been added as a favorite command.
The Tilt Angle controls the strength and direction of the horizontal shear.
An angle of 0 degrees produces no geometric change.
Positive and negative values apply shear in opposite directions.
Adjust the angle until structures that should be vertical or geometrically aligned appear correctly positioned in the preview.
The preview allows the geometric effect to be evaluated before modifying the complete hyperspectral cube.
The preview is generated from the average image across the spectral bands.
Changing the angle updates the preview without modifying the working hyperspectral dataset.
The full hyperspectral cube is transformed only after Apply is selected.
Geometric transformations usually place output pixels between original pixel locations.
Interpolation determines how intensity values at these new locations are calculated.
Bilinear interpolation is recommended for most continuous hyperspectral intensity images.
It calculates new pixel intensities from nearby pixels and generally provides a good balance between smoothness and preservation of image detail.
Nearest interpolation assigns the value of the nearest original pixel.
It avoids averaging neighboring values but can produce a more block-like appearance.
Nearest interpolation can be useful when preserving discrete pixel values is more important than obtaining a smooth image.
Bicubic interpolation uses a larger local neighborhood and generally produces a smoother result.
It can modify intensity values more strongly than Nearest or Bilinear interpolation and should therefore be selected with consideration for quantitative data.
Output - Preserve Original Size
Preserve Original Size keeps the spatial dimensions of the hyperspectral cube unchanged.
For example, a 640 x 480 image remains 640 x 480 after the transformation.
Because the output canvas does not expand, portions of the sheared image that move outside the original image boundary can be clipped.
Use this option when maintaining identical spatial dimensions is important.
Expand to Fit enlarges the output canvas so that the complete transformed image can be retained.
The number of rows or columns may therefore increase.
Use this option when preserving the entire transformed field of view is more important than maintaining the original spatial dimensions.
A shear transformation can create output pixels for which no corresponding original image data exist.
The Fill Value determines the value assigned to these newly created regions.
A Fill Value of 0 is appropriate for many datasets, but the appropriate value depends on the intensity scale and interpretation of the hyperspectral data.
Filled regions do not represent measured spectral information and should be interpreted accordingly.
1. Open Tilt / Shear Correction.
4. Select the desired interpolation method.
5. Choose Preserve Original Size or Expand to Fit.
6. Select an appropriate Fill Value.
IDCubePro then applies the same transformation to every spectral band in the current working cube.
Every spectral band is transformed using the same geometric transformation and output geometry.
This is important because hyperspectral analysis depends on corresponding spatial pixels remaining aligned across wavelengths.
The operation changes spatial sampling but does not intentionally shift one spectral band relative to another.
Tilt / Shear Correction is a geometric resampling operation.
Except when Nearest interpolation is used, transformed pixel intensities can be calculated from combinations of neighboring original pixels.
Therefore, shear correction should be used when geometric correction is scientifically justified rather than solely to improve image appearance.
For quantitative analysis, record the transformation settings and interpolation method used.
When Preserve Original Size is selected, parts of the transformed image can move outside the available image area.
Those regions are not retained.
If important sample information approaches the image boundaries, consider using Expand to Fit.
Expand to Fit can introduce regions outside the original measured field of view.
These regions contain the selected Fill Value rather than measured hyperspectral information.
They should not normally be interpreted as valid sample measurements.
The Tilt / Shear operation modifies the current working hyperspectral dataset.
The original dataset is preserved by IDCubePro.
Use Reset Image when you want to return to the original image data and remove the applied geometric processing.
The complete image has the wrong orientation.
A horizontal feature should become vertical.
The entire field of view needs to be turned by a fixed angle.
Use Tilt / Shear Correction when:
The overall orientation is correct but structures appear geometrically slanted.
Different image rows require progressive horizontal displacement.
You need to correct systematic horizontal skew.
Flip Horizontal reverses the image from left to right.
Flip Vertical reverses the image from top to bottom.
Neither operation corrects geometric skew.
Use Tilt / Shear Correction when progressive displacement rather than image reversal is required.
TILT / SHEAR OR SPATIAL CROP?
Spatial Crop removes unwanted rows and columns from the field of view.
It does not correct geometric skew.
You can apply Tilt / Shear Correction and Spatial Crop sequentially when both geometric correction and removal of unwanted image boundaries are required.
1. Confirm that the correct hyperspectral dataset is loaded.
2. Determine whether the problem is rotation or geometric shear.
3. Open Tilt / Shear Correction.
4. Begin with a small correction angle.
5. Adjust the angle while inspecting structures with known geometry.
6. Use Bilinear interpolation as a general starting point for continuous intensity data.
7. Use Expand to Fit if clipping would remove important image information.
9. Inspect the resulting image and image boundaries.
10. Reset and repeat with a different angle if the correction is unsatisfactory.
Tilt / Shear Correction cannot reconstruct spatial information that was not measured during acquisition.
Interpolation creates resampled intensity values but does not create new measured spectral information.
Large shear angles cause greater spatial displacement and may require more interpolation.
Filled areas introduced by the transformation contain no original measured information.
Geometric correction should therefore be applied conservatively and documented when used for quantitative analysis.
Rotate 90 degrees Counterclockwise Close Tilt / Shear Correction help.