correlation notebook and start of formatting of slides

config.toml

@@ -2,5 +2,7 @@ baseURL = "https://bellonet.github.io/image-registration-workshop/"
 languageCode = "en-us"
 title = "Image Registration Workshop"
 
-[markup.goldmark.renderer]
-  unsafe = true
+[markup]
+  [markup.goldmark]
+    [markup.goldmark.renderer]
+      unsafe = true

content/_index.md

@@ -1,8 +1,10 @@
 ---
 title: "Image Registration Workshop"
-date: 2024-09-20
+date: 2024-09-23
 draft: false
 layout: workshop
+author: Ella Bahry
+cover: img/registration.png
 ---
 
 ## What is Image Registration?
@@ -11,92 +13,113 @@ layout: workshop
 In this workshop, we will explore fundamental concepts and practical techniques for image registration, focusing on applications in microscopy, material science, and earth science.
 {{</ notes >}}
 
-{{< horizontal >}}
-
-<span style="font-size: 24px; color: white;">♡</span>
-<span style="font-size: 24px; color: white;">♡</span>
-
-<img src="img/registration.png" alt="MRI CT registration" style="width: 700px;" />
-
-<span style="font-size: 24px; color: white;">♡</span>
-<span style="font-size: 24px; color: white;">♡</span>
-
-{{</ horizontal >}}
+![](img/registration_big.png)
 
 <h4>Spatial alignment of two or more images.</h4>
+
+- In science, it's an essential  step for comparing or integrating data
 
-{{< notes >}}
-Image registration is the process of transforming different sets of data into one coordinate system.
-{{</ notes >}}
-
----
-
-## Applications in Various Fields
 
 {{< notes >}}
-Image registration is widely used across multiple disciplines.
+Image registration is the process of aligning multiple datasets into a common coordinate system, enabling accurate comparison and analysis.
 {{</ notes >}}
 
-- **Microscopy**: Aligning slices in a 3D stack, channels, runs, tiles, modalities, or time points.
-- **Medical Imaging**: Viewpoints, stacks, normalization to an atlas, co-registering images from different modalities (e.g., MRI, CT).
-- **Material Science**: Comparing material properties under varying conditions.
-- **Earth Science**: Aligning satellite images for change detection, georeferencing, integration from different sensors.
-
 ---
 
 ## Common needs in research
 
 {{< horizontal >}}
-
 Multimodal:
 ![](img/multimodal.jpg)
 
+Stitching:
+![](img/stitching.jpg)
+
 Stack:
 ![](img/slice_to_slice.jpg)  
 
 Viewpoint:
 ![](img/coregistration.jpg)
 
-Stitching:
-![](img/stitching.jpg)
-
 Temporal:
 ![](img/timesteps.jpg)
 
 {{</ horizontal >}}
 
-- Essential for comparing or integrating data
+---
+
+## Applications in Various Fields
+
+{{< notes >}}
+Image registration is widely used across multiple disciplines.
+{{</ notes >}}
+
+- **Microscopy**: Aligning slices in a 3D stack, channels, runs, time points, tiles (stitching), and modalities.
+<br><br>
+- **Medical Imaging**: Viewpoints, stacks, normalization to an atlas, co-registering images from different modalities (e.g., MRI, CT).
+<br><br>
+- **Material Science**: Comparing material properties under varying conditions.
+<br><br>
+- **Earth Science**: Georeferencing, integration from different sensors, aligning satellite images for change detection.
 
 ---
 
 ## Integrated Techniques: One-Step Registration
 
 {{< notes >}}
-Introducing registration methods that integrate detection and transformation into a single process.
+Introducing registration methods that combine both matching and transformation into one smooth process, simplifying image alignment. 
 {{</ notes >}}
 
 - **Intensity-Based Registration**
-  - Iterative - Optimize pixel intensity similarities directly to align images
+  - Iterative process that optimizes aligned pixel intensity similarities (e.g. **correlation coefficient** or **MSE**).
 - **Mutual Information-Based Registration**
-  - Iterative - Use statistical dependence to align multimodal images
+  - Iteratively aligns multimodal images by maximizing the statistical relationship between them.
 - **Frequency Domain Methods**
-  - Compute transformations using Fourier transforms in the frequency domain.
+  - Transforms images into the Fourier space to compute alignment transformations.
 - **Deep Learning-Based Registration**
-  - Neural networks predict transformations directly from image data
+  - Uses neural networks to predict transformations from image data, learning complex patterns.
+
+---
+
+## Example Technique: Correlation Coefficient
+
+![](img/correlation_r45_s1.5.png)
+![](img/correlation_r10_s1.1.png)
+![](img/correlation_r2_s1.png)
+![](img/correlation_r0_s1.png)
+
+{{< notes >}}
+[https://github.com/bellonet/image-registration-workshop/blob/main/example_notebooks/correlation_example.ipynb
+](https://github.com/bellonet/image-registration-workshop/blob/main/example_notebooks/correlation_example.ipynb
+)
+{{</ notes >}}
 
 ---
 
 ## Example Technique: Mutual Information
 
+{{< horizontal >}}
+
 <img src="img/joint_histogram_iterative.gif" alt="Mutual Information iterations"/>
 
+![](img/mutual_information_equation.png)
+
+{{</ horizontal >}}
+
+- Looking at the equation, the more structure we have in the joint histogram, the lower its entropy, and thus the mutual information is higher.
+
+{{< notes >}}
+Entropy is maximized when there is maximum uncertainty or randomness in the pixel intensities.  
+Meaning that an image with a single pixel intensity value will have minimum entropy, and an image with a uniform distribution of pixel intensities will have maximum entropy.
+{{</ notes >}}
+
 - [https://github.com/bellonet/image-registration-workshop/blob/main/example_notebooks/mutual_information.ipynb
 ](https://github.com/bellonet/image-registration-workshop/blob/main/example_notebooks/mutual_information.ipynb
 ) 
 
 ---
 
-## 2 Step Techniques - Feature Detection & Transformation
+<h2>2 Step Techniques - Feature Detection & Transformation</h2>
 
 {{< notes >}}
  An overview of common methods used in image registration, highlighting the two main steps involved.

static/css/main.css

@@ -425,7 +425,7 @@ section .horizontal > figure figcaption {
 
 section .presentation-title {
     flex: 1;
-    width: 50%;
+    width: 30%;
     display: flex;
     flex-direction: column;
     justify-content: space-evenly;

static/img/edit_img.ipynb

@@ -0,0 +1,125 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "id": "7f71970e-f2f7-48ac-b9f1-507dcc677c5c",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from PIL import Image, ImageOps\n",
+    "import os"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 10,
+   "id": "23c64cd7-1ee0-4eef-b1a8-30e83e8384ef",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def save_img_copy(im_path):\n",
+    "    # Open the image\n",
+    "    img = Image.open(im_path)\n",
+    "\n",
+    "    # Save the original image as f'{im_path}1'\n",
+    "    base, ext = os.path.splitext(im_path)\n",
+    "    original_save_path = f'{base}1{ext}'\n",
+    "    img.save(original_save_path)\n",
+    "    return img"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 21,
+   "id": "9f5e11b0-5a4c-4760-98f0-05cabafd33ef",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def resize_image(im_path, resizing_factor):\n",
+    "\n",
+    "    img = save_img_copy(im_path)\n",
+    "\n",
+    "    new_size = (int(img.width * resizing_factor), int(img.height * resizing_factor))\n",
+    "\n",
+    "    if resizing_factor < 1.0:\n",
+    "        # Resize with antialias filter when downsampling\n",
+    "        resized_img = img.resize(new_size, Image.ANTIALIAS)\n",
+    "    else:\n",
+    "        # Just copy the image if upsampling or maintaining size\n",
+    "        resized_img = img.resize(new_size)\n",
+    "\n",
+    "    # Save the resized image to the original path\n",
+    "    resized_img.save(im_path)\n",
+    "    "
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 22,
+   "id": "7b151e2c-5d7a-43fb-992c-ac6ad2e85091",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def pad_image(im_path, padding):\n",
+    "\n",
+    "    img = save_img_copy(im_path)\n",
+    "    \n",
+    "    # Apply the padding (black padding by default)\n",
+    "    padded_img = ImageOps.expand(img, border=padding, fill=(0, 0, 0))  # black fill\n",
+    "    \n",
+    "    # Save the padded image to the specified path\n",
+    "    padded_img.save(im_path)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 23,
+   "id": "5b517f7e-99ba-4c03-9f5f-acb9cebfe86a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "pad_image('registration.png', 100)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "id": "c1c37b68-a636-4442-a877-a714f169eacc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "resize_image('registration_big.png', 1.5)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1e3fffdb-1355-4790-940f-56e84abd0f5f",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.10.12"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

static/reveal/dist/theme/white-hi.css

@@ -67,6 +67,7 @@ section.has-dark-background, section.has-dark-background h1, section.has-dark-ba
 .reveal .slides {
     display: flex;
     text-align: left;
+    flex: 1;
 }
 
 .reveal .slides section,
@@ -116,6 +117,7 @@ section.subsection h1, section.subsection h2 {
 
 .reveal .toc {
     flex: 1;
+    font-size: 2vh;
 }
 
 .reveal .qr-code-slides {