BIOL643 Lecture 3 Image Processing and Manipulation UBt PDF
Document Details
Uploaded by FantasticCyan8252
Noura S. Abou Zeinab
Tags
Summary
This document presents a lecture on image processing techniques, covering methods like thresholding, gradient ascent/descent, watershed segmentation, and edge detection. It also discusses the difference between image processing and manipulation. Image processing is presented as scientifically valuable whereas image manipulation is scientifically unacceptable.
Full Transcript
Dr. Noura S. Abou Zeinab Assistant Professor in Cell Biology and Histology Segmentation automatically identifies and separates distinct objects within an image to analyze them further. Specifically, it divides an image into How it works: It uses a defined threshold intensity. All pixels above...
Dr. Noura S. Abou Zeinab Assistant Professor in Cell Biology and Histology Segmentation automatically identifies and separates distinct objects within an image to analyze them further. Specifically, it divides an image into How it works: It uses a defined threshold intensity. All pixels above this threshold are marked as objects, and those below are marked as the background. Use case: It’s often used in simpler cases, like background extraction. Limitations: It struggles when different objects have varying intensities. Key tool: In Fiji (image processing software), it’s done using Image>Adjust>Threshold. How it works: This technique finds local intensity maxima (or minima) by treating an image like a landscape of peaks and valleys. The algorithm tracks from each pixel to its neighboring higher-intensity pixel until it reaches the maximum. Use case: Best for identifying objects that don’t have the same intensity across the image. Key concept: The “Basin of Attraction” (BOA) is used, which groups points that all lead to the same local maximum. How it works: Similar to how water fills basins, this method seeds each cell, and pixels are grouped by flooding areas of intensity. Each “cell” is filled, like pouring water into the intensity landscape. Use case: Ideal for identifying individual objects in crowded or overlapping regions (e.g., cells). Key tool: It’s demonstrated using MorphoGraphX software. How it works: It identifies the edges of objects by detecting sudden changes in intensity across the image. Use case: Useful for detecting object boundaries or outlines. Key tool: Fiji’s Process>Find edges tool or the Canny edge detection plugin. II- Electron Microscopy Image Processing V.S. Manipulation
