Skip to content

Image processing

This tutorial will walk you through the steps of preprocessing, segmenting, and aligning images using the Streamlit application. These steps are designed to prepare your images for further analysis by standardizing their visual properties, extracting key regions of interest, and aligning them to a base image. Finally, we will focus on inspecting the heatmap output of the alignment step to uncover structural similarities between images.

Step 1: Preprocessing Images

Overview

Preprocessing involves enhancing and standardizing the images to make them more suitable for subsequent analysis. This step typically includes adjustments to the images' brightness, contrast, and sharpness.

Instructions

  1. Upload Your Images: Begin by uploading the images you intend to process.
  2. Equalize Brightness: Adjust the brightness across all images to a common level to eliminate variations caused by different lighting conditions.
  3. Equalize Contrast: Standardize the contrast to ensure that the differences between the light and dark regions are consistent.
  4. Equalize Sharpness: Enhance the sharpness uniformly across all images, making fine details clearer and more comparable.

The preprocessing step helps ensure that all images share consistent visual properties, setting the stage for accurate segmentation and alignment.

Step 2: Segmenting Images

Overview

Segmentation involves isolating the key regions of interest within each image. This step is crucial for focusing on the parts of the image that are most relevant for further analysis, such as specific patterns or objects.

Instructions

  1. Choose a Segmentation Method: Use the application to select a segmentation method appropriate for your images. For this tutorial, let's choose the Thresholding method, which isolates regions of the image based on pixel intensity.
  2. Set Threshold Levels: Adjust the threshold levels to define the regions of interest clearly. The goal is to isolate the key structural components while minimizing background noise.
  3. Inspect Segmented Images: Once segmentation is complete, review the output to ensure that the relevant regions have been accurately isolated.

Segmentation sharpens the focus on the parts of the image that are most likely to be informative in later analysis, such as alignment.

Step 3: Aligning Images to a Base Image

Overview

Alignment is the process of adjusting all images to match a base image in terms of position, orientation, and scale. This step ensures that all images are consistent with each other, which is essential for uncovering structural similarities.

Instructions

  1. Select a Base Image: Choose one image from your set to serve as the base image. This image will be used as the reference for aligning all other images.
  2. Choose a Motion Model: Select the Affine motion model. This model allows for rotation, scaling, translation, and skewing, offering the flexibility needed to align images with different perspectives or distortions.
  3. Align Images: Align each image in your set to the base image using the selected motion model.

Inspecting the Alignment Heatmap

After alignment, the application will generate a heatmap that visualizes the similarities between the aligned images and the base image. The heatmap provides a powerful tool for uncovering structural similarities that might not be immediately obvious.

  1. Review the Heatmap: Examine the heatmap closely. Areas of high similarity will appear as clusters of warm colors (reds and oranges), while areas of lower similarity will appear cooler (blues and greens).
  2. Identify Similarities: Use the heatmap to identify regions where the images closely match the base image. These areas of high similarity can indicate shared structural features, such as common patterns, textures, or shapes.
  3. Analyze Differences: Conversely, areas of low similarity might highlight unique features or variations between the images, which could be of interest depending on the focus of your analysis.

Conclusion

By following these steps—preprocessing, segmenting, and aligning images—you can effectively prepare your images for detailed analysis. The heatmap generated during the alignment step is particularly valuable for uncovering structural similarities between images, offering insights that can inform your research or analysis. Whether you're working with art history images or other types of visual data, this process will help you standardize and compare your images systematically.