Computer Vision Overview

Computer vision aims to enable computers to 'see' and interpret images and videos in the same way humans do, understanding objects, scenes, and actions.

The process of recovering the 3D shape and appearance of objects from images or videos. This is a challenging task due to the complex nature of the visual world.

Finding corresponding points between two images of the same scene, taken from slightly different viewpoints. This is crucial for reconstructing 3D models.

Computer vision models that learn from large datasets of images and labels, allowing them to identify patterns and make predictions. This is a powerful technique for tasks like object recognition, image classification, and scene understanding.

A problem where we try to find a solution based on limited information. In vision, we try to understand the world from the incomplete information in an image.

Techniques based on physics and probability to make sense of the visual world, helping computers understand the relationships between objects and their properties.

Despite advances in computer vision, accurately understanding an image at the level of a young child remains a challenge. This is due to the complexity of the visual world and the limitations of current techniques.

The human visual system is incredibly complex and still not fully understood. This complexity makes it difficult to replicate human vision using computers.

The process of modeling how objects appear in images, taking into account factors like light, motion, and camera properties.

Branches of physics that study how light interacts with objects and the principles of image formation.

Computer Graphics and Forward Models in Computer Vision are interconnected, both aim to create realistic images by simulating how light interacts with objects and environments.

The phenomenon where the visual system corrects for variations in illumination when interpreting the color of objects.

A visual illusion that demonstrates the visual system's sensitivity to perceived perspective and depth cues.

A visual illusion illustrating how the visual system interprets the brightness of an object relative to its surroundings.

The ability of the visual system to perceive and interpret light, color, and depth cues to understand the world around us.

The way objects move, change shape, and interact with each other in a scene.

Combining multiple overlapping photos to create a seamless panoramic image.

Merging multiple photos taken with different exposures to create a perfectly exposed image.

Transforming one image into another through a smooth transition. This is often used to create dynamic effects.

Creating a 3D model of an object or person from one or more photos.

The process of tracking points in a video to insert 2D images or 3D models. Also used to stabilize shaky footage.

Navigating through a collection of photos, often representing a space like a house, by flying between them in a 3D environment.

A computer vision technique that identifies faces within an image.

Using images for automatic authentication, often by recognizing faces for access control.

The process of using images to create realistic 3D models, commonly used in fields like computer graphics and animation.

A technique for automatically stitching multiple images together to create a larger, panoramic view.

The field of computer vision that focuses on using computer vision techniques to enhance or manipulate images, often with the goal of producing improved or creative results.

Techniques used to capture and process images with a wide range of brightness values, allowing for details in both bright and dark areas.

Algorithms that convert images with a wide dynamic range (HDR) to a displayable range, making them visible on monitors or prints.

A computer vision technique that analyzes images to recognize objects based on their distinctive features and shapes.

A model for object recognition that focuses on identifying constellations of distinctive features within an image.

A model for object recognition that represents objects as a set of interconnected parts or features, allowing for more complex descriptions.

A color space represented by three components: Y (luminance), Cb (blue chroma difference), and Cr (red chroma difference). It's often used in image compression standards like JPEG.

The RGB color values obtained from a JPEG image are actually gamma-compressed. This means they are not the raw values, but rather a non-linear representation.

A color space where colors are represented by their hue, saturation, and value. It's useful for tasks like color picking or manipulating an image's overall color appearance.

A technique used in image compression to reduce the amount of data needed to store an image. It involves removing redundant information and representing color differences more efficiently.

A type of image artifact that can appear during image compression, particularly in JPEGs, where sharp edges or block boundaries become visible.

A color space where colors are represented as normalized ratios of red, green, and blue components. It simplifies color manipulation by focusing on relative color proportions.

The process of restoring the original color information in an image by adjusting for gamma compression. This can be important when working with images that have been compressed using JPEG.

A technique that involves analyzing and potentially correcting image artifacts that arise from compression, especially blockiness in images.

A method in image compression where the image is broken down into 8x8 blocks, transformed using Discrete Cosine Transform (DCT), and coefficients are quantized for efficient storage.

A measure of the amount of detail preserved in a compressed image. Higher quality settings result in more detail, larger file size.

A technique used in video compression to predict the motion of objects in a sequence of frames. It only encodes the changes between frames, saving space.

A standard for compressing video and audio, widely used in streaming services and DVDs. Combines DCT transform, motion compensation, and other techniques for efficient encoding.

A technique used in data compression, like JPEG, where the lowest frequency coefficient is used to predict the values of other coefficients, reducing redundancy.

JPEG 2000 and JPEG XR use advanced techniques such as wavelets and lapped transforms to improve compression efficiency compared to traditional JPEG.

A standard for video compression using only individual JPEG compressed frames, without motion compensation, resulting in larger file sizes.

A new open standard for video compression that uses smaller blocks (4x4 or 2x2) and advanced techniques for better quality and efficiency.