Computer Vision Lecture Notes PDF

Computer Vision By Dr. Ahmed Taha Lecturer, Computer Science Department, Faculty of Computers & Artificial Intelligence, Benha University 1  Introduction Lecture One 2 3 4 What is Compute...

Computer Vision By Dr. Ahmed Taha Lecturer, Computer Science Department, Faculty of Computers & Artificial Intelligence, Benha University 1  Introduction Lecture One 2 3 4 What is Computer Vision? A process that produces from images of the external world a description that is useful to the viewer and not cluttered with irrelevant information (Marr) Construction of explicit , meaningful descriptions of physical objects from images (Ballard and Brown ) To make useful decisions about real physical objects and scenes based on sensed images (Shapiro and Stockman) 5 Related Fields - Image Processing 6 Related Fields - Image Processing Image Processing: ▪ Image in, image out ▪ Usually low level techniques ( eg , compression , edge) ▪ Quantitative measurements Computer Vision: ▪ Extracting symbolic descriptions ▪ Higher level techniques ( eg , object recognition) ▪ Semantic (quantitative or qualitative ) output Image processing techniques are often used in computer vision The definitions are not firm ‐ there is overlap between the fields 7 Related Fields Pattern recognition: ▪ Recognition of patterns (classification) ▪ Inputs often represented as feature vectors ▪ Techniques useful for 2D and constrained 3D image recognition problems , but usually too limited for general 3D problems 8  Related Fields Photogrammetry: ▪ Concerned with accurately measuring properties from images ▪ An older field ‐ historically focused on remote sensing (e.g., images from airplanes or satellites) ▪ Computer vision concerned with more than just measuring ▪ However , many techniques are the same or similar 9 Related Fields – Computer Graphics Computer vision is the inverse of computer Graphics Computer Graphics 3D Models of objects, locations Lighting Images information Camera parameters Computer Vision The forward process is unique, the inverse process is not! 10 11 12 13 Computer Vision 14 What does WALL-E see? 15 Low-Level: Resizing 16 Low-Level: Resizing Nearest Neighbor Bilinear 17 Low-Level: Image Adjustments 18 Low-Level: Grayscale 19 Low-Level: Exposure 20 Low-Level: Saturation 21 Low-Level: Hue 22 Low-Level: Edges 23 Low-Level: Oriented Gradients 24 Low-Level: Oriented Gradients 25 Low-Level: Segmentation (color) 26 Low-Level Vision Photo manipulation - Size - Color - Exposure - X-Pro II Feature extraction - Edges - Oriented gradients - Segments Low level vision is exciting!!! #latergram #nofilter 27 Low-Level Vision Applications? Anyone? 28 Mid-Level Vision 29 Mid-Level: Panorama Stitching 30 Mid-Level: Panorama Stitching 31 Mid-Level: Panorama Stitching 32 Mid-Level: Multi-View Stereo 33 Mid-Level: Multi-View Stereo 34 Mid-Level: Multi-View Stereo 35 Mid-Level: Multi-View Stereo 36 Mid-Level: Structured Light Scan 37 Mid-Level: Structured Light Scan 38 Mid-Level: Range Finding 39 Mid-Level: Optical Flow 40 Mid-Level: Optical Flow 41 Mid-Level: Time Lapse 42 Mid-Level Vision Image Image - Panoramas Image World - Multi-view stereo - Structure from motion - Structured light - LIDAR Image Time - Optical flow - Time lapse 43 High-Level Vision 44 High-Level: Classification - What is in the image? 45 High-Level: Tagging - What are ALL the things in the image? 46 High-Level: Detection - What are ALL the things in the image? - Where are the? 47 High-Level: Semantic Segmentation 48 High-Level: Instance Segmentation 49 High-Level: So many other things - Single image 3D - Game playing - Super-resolution - Retrieval - Other cool things, yay! 50 High-Level Vision Semantics! - Image classification - Object detection - Segmentation Applications - Retrieval - Robots? - and…???? 51 52