Stitched with feather blend

Oct 13, 2025 by wbrum.org 121 views

Introduction

Feather blending is a popular method used in image stitching to merge multiple images into a single, seamless image. However, in some cases, feather blending can produce unexpected results, such as a completely black (0 intensity) overlapped region. In this article, we will explore the issue of feather blending generating a stitched image with 0 intensity in the overlapped region and provide a solution to this problem.

Background

Image stitching is a technique used to combine multiple images taken from different viewpoints into a single, panoramic image. The process involves aligning the images, removing any overlapping regions, and merging the images into a single image. Feather blending is a method used to merge the images, where the intensity of the pixels in the overlapped region is blended together to create a seamless image.

The Problem

When using the XR_matlab_stitching_wrapper() function with BlendMethod = 'feather', the code works well on demo lightsheet datasets. However, when migrating to work on our own lightsheet dataset collected from a similar openSPIM setting, setting BlendMethod = 'feather' led to a stitched image with a completely black (0 intensity) overlapped region. This issue was not observed when changing BlendMethod to 'mean' without changing any other settings.

Settings

The settings used to produce the completely black overlapped region were:

DS = false;
DSR = false;
SkewAngle = 39.4;
Reverse = false;
xyPixelSize = 4.6/16*2;
dz = 1.4;
parseSettingFile = false;
axisOrder = 'x,y,z';
BlendMethod = 'feather';
overlapType = 'full';
stitchResultDir = 'matlab_stitch';
xcorrShift = true;
PrimaryCh = 'CamA_ch0';
ChannelPatterns = {'CamA_ch0'};
zarrFile = true;
largeFile = false;
poolSize = [2,2,2];
Save16bit = true;
xcorrThresh = 0.15;

Stitched Images

The following images show the stitched images produced using feather blending and mean blending:

Stitched with feather blend
Stitched with mean blend

Solution

To solve the issue of feather blending generating a stitched image with 0 intensity in the overlapped region, we need to adjust the settings used in the XR_matlab_stitching_wrapper() function. Specifically, we need to change the BlendMethod setting from 'feather' to 'mean'. This will ensure that the intensity of the pixels in the overlapped region is calculated using the mean value, rather than the feather blending method.

Code Changes

To implement the solution, we need to modify the code used in the XR_matlab_stitching_wrapper() function. Specifically, we need to change the BlendMethod setting from 'feather' to 'mean'. The modified code is shown below:

BlendMethod = 'mean';

Conclusion

In conclusion, feather blending can produce unexpected results, such as a completely black (0 intensity) overlapped region. However, by changing the BlendMethod setting from 'feather' to 'mean', we can solve this issue and produce a stitched image with a seamless overlapped region. We hope that this article has provided a useful solution to this problem and has helped to improve the quality of image stitching results.

Future Work

Future work could involve investigating other methods for merging images, such as the use of deep learning techniques. Additionally, we could explore the use of other settings, such as the xcorrThresh setting, to improve the quality of the stitched image.

References

[1] XR_matlab_stitching_wrapper() function documentation
[2] Image stitching tutorial
[3] Feather blending method documentation

Appendix

The following appendix provides additional information on the settings used in the XR_matlab_stitching_wrapper() function.

Settings

The following settings were used in the XR_matlab_stitching_wrapper() function:

DS = false;
DSR = false;
SkewAngle = 39.4;
Reverse = false;
xyPixelSize = 4.6/16*2;
dz = 1.4;
parseSettingFile = false;
axisOrder = 'x,y,z';
BlendMethod = 'feather';
overlapType = 'full';
stitchResultDir = 'matlab_stitch';
xcorrShift = true;
PrimaryCh = 'CamA_ch0';
ChannelPatterns = {'CamA_ch0'};
zarrFile = true;
largeFile = false;
poolSize = [2,2,2];
Save16bit = true;
xcorrThresh = 0.15;

Stitched Images

The following images show the stitched images produced using feather blending and mean blending:

Stitched with feather blend
Stitched with mean blend Feather Blending Generates Stitched Image with 0 Intensity in Overlapped Region: Q&A ====================================================================================

Introduction

In our previous article, we discussed the issue of feather blending generating a stitched image with 0 intensity in the overlapped region. We also provided a solution to this problem by changing the BlendMethod setting from 'feather' to 'mean'. In this article, we will provide a Q&A section to address any questions or concerns that readers may have.

Q: What is feather blending and how does it work?

A: Feather blending is a method used in image stitching to merge multiple images into a single, seamless image. It works by blending the intensity of the pixels in the overlapped region together to create a seamless image.

Q: Why does feather blending produce a stitched image with 0 intensity in the overlapped region?

A: Feather blending can produce a stitched image with 0 intensity in the overlapped region due to the way it blends the intensity of the pixels together. When the intensity of the pixels in the overlapped region is blended together, it can result in a value of 0, which can appear as a black or dark region in the stitched image.

Q: How can I prevent feather blending from producing a stitched image with 0 intensity in the overlapped region?

A: To prevent feather blending from producing a stitched image with 0 intensity in the overlapped region, you can change the BlendMethod setting from 'feather' to 'mean'. This will ensure that the intensity of the pixels in the overlapped region is calculated using the mean value, rather than the feather blending method.

Q: What are the benefits of using mean blending instead of feather blending?

A: The benefits of using mean blending instead of feather blending include:

Reduced risk of producing a stitched image with 0 intensity in the overlapped region
Improved quality of the stitched image
Easier to adjust the blending method to achieve the desired result

Q: Can I use other blending methods instead of mean blending?

A: Yes, you can use other blending methods instead of mean blending. Some popular blending methods include:

Weighted average: This method calculates the intensity of the pixels in the overlapped region by taking a weighted average of the intensities of the individual pixels.
Maximum: This method calculates the intensity of the pixels in the overlapped region by taking the maximum value of the intensities of the individual pixels.
Minimum: This method calculates the intensity of the pixels in the overlapped region by taking the minimum value of the intensities of the individual pixels.

Q: How do I adjust the blending method to achieve the desired result?

A: To adjust the blending method to achieve the desired result, you can experiment with different blending methods and adjust the parameters of the blending method to achieve the desired result. For example, you can adjust the weight of the weighted average method or the threshold of the maximum method to achieve the desired result.

Q: Can I use feather blending with other image stitching methods?

A: Yes, you can use feather blending with other image stitching methods. However, you may need to adjust the parameters of the feather blending method to achieve the desired result.

Q: What are the limitations of feather blending?

A: The limitations of feather blending include:

Risk of producing a stitched image with 0 intensity in the overlapped region
Difficulty in adjusting the blending method to achieve the desired result
Limited flexibility in terms of adjusting the parameters of the blending method

Conclusion

In conclusion, feather blending can produce a stitched image with 0 intensity in the overlapped region due to the way it blends the intensity of the pixels together. However, by changing the BlendMethod setting from 'feather' to 'mean', we can solve this issue and produce a stitched image with a seamless overlapped region. We hope that this Q&A article has provided a useful resource for readers who have questions or concerns about feather blending and image stitching.

Future Work

Future work could involve investigating other blending methods and adjusting the parameters of the blending method to achieve the desired result. Additionally, we could explore the use of other image stitching methods and adjusting the parameters of the image stitching method to achieve the desired result.

References

[1] XR_matlab_stitching_wrapper() function documentation
[2] Image stitching tutorial
[3] Feather blending method documentation

Appendix

The following appendix provides additional information on the settings used in the XR_matlab_stitching_wrapper() function.

Settings

The following settings were used in the XR_matlab_stitching_wrapper() function:

DS = false;
DSR = false;
SkewAngle = 39.4;
Reverse = false;
xyPixelSize = 4.6/16*2;
dz = 1.4;
parseSettingFile = false;
axisOrder = 'x,y,z';
BlendMethod = 'feather';
overlapType = 'full';
stitchResultDir = 'matlab_stitch';
xcorrShift = true;
PrimaryCh = 'CamA_ch0';
ChannelPatterns = {'CamA_ch0'};
zarrFile = true;
largeFile = false;
poolSize = [2,2,2];
Save16bit = true;
xcorrThresh = 0.15;

Stitched Images

The following images show the stitched images produced using feather blending and mean blending: