Advanced Multimedia Lecture 4

Study Notes

Computer Imaging: Defined as the acquisition and processing of visual information by a computer. The ultimate receiver of this information is either the computer or the human visual system. This leads to two categories: computer vision and image processing.

Image Analysis: A subfield of computer vision that examines image data to solve vision problems. Key topics include:
- Feature Extraction: Acquiring higher-level information from the image.
- Pattern Classification: Using the higher-level information to identify objects in the image.

Sample per point (B&W or Gray scale): A grayscale image, one value per pixel
Samples per point (RGB): An RGB image has three values per pixel (Red, Green, Blue)
Samples per point (RGBA): Similar to RGB, but with an additional "Alpha" channel (RGBA) for opacity control.

Color Image: The number of color values in a color image is 2²⁴ (16,777,216 colors).
RGB Components: Each pixel in a color image has three values representing Red, Green, and Blue. Combining these values creates a new color.
Color Value Range: Each color component (Red, Green, Blue) has values ranging from 0 to 255. This value range determines the color's intensity (lightness or darkness).
3D Representation: A colour image is represented as a 3D matrix with the rows and columns representing the horizontal and vertical pixel locations and the third dimension defining the RGB values.
Examples of Pixel Values: [(90, 0, 53)], [(213, 60, 67)], [(249, 215, 203)] represent different color intensities at specific pixel locations

Bit Depth and Number of Colors: The table links the number of bits per pixel to the number of possible colors or shades in an image.
Image Size Calculation: The calculation for the file size (in KB) in a RAW bitmap file is: (M * N * B) / (8 * 1024), where M & N = horizontal and vertical resolution, B = number of bits per pixel.

Two Major Tasks:
- Improvement of Pictorial Information for Human Interpretation: Enhancing the quality or details in images for easier human understanding (e.g., medical imaging, photo enhancement).
- Processing of Image Data for Storage, Transmission and Representation for Machine Perception: Data processing for machines, such as storing and transmitting the data in a way that machines can understand.

Low Level Processes: Processes that involve changes to pixel values to detect anomalies like noises. Examples: noise removal and image sharpening. Input: Image, Output: Image
Mid Level Processes: Processes that manipulate attributes to solve problems and include image segmentation and object recognition. Input: Image, Output: Attributes.
High Level Processes: Processes that involve higher-level tasks/understanding such as recognizing scenes and doing object detection. Input: Attributes, Output: Understanding.

Image Enhancement: Techniques used to improve or highlight features in an image.
Hubble Telescope: Image processing techniques were used to correct for errors in the original Hubble Telescope mirror.
Medicine): Using medical images (e.g., MRI scans) and image processing to find and highlight tissue boundaries

Raw Storage: Simple byte-by-byte storage format for bitmaps, often called RAW files..
File Size Formula: File size (KB) = (M * N * B) / (8 * 1024)

Image Acquisition (input) → Preprocessing (image restoration, morphological processing,…) → Image Enhancement → Color Image Processing → Image Compression → Segmentation → Object recognition → Representation & Description (Feature Selection)

Digital Image Acquisition: Images may be acquired already in digital format thus involving basic preprocessing like scaling

Image Enhancement deals with simple subjective improvements to the image
Image Restoration deals with objective improvements to the image using mathematical/probabilistic models
Morphological Processing: Deals with tools for extracting image components useful in the representation and description of shape.

Representation and Description (Feature Selection): Extracting attributes resulting in quantitative information to differentiate object classes

Compression: Techniques for reducing storage/bandwidth for images (e.g., JPEG)

Importance: Gaining importance due to the increased use of digital images.
Color Information: Utilizes color information to extract relevant elements and features from images.