Computer Vision

Name 5 parts of the eye

1. Cornea 2. Iris 3. Pupil 4. Lens 5. Retina.

Why do we apply grayscale to an image when we apply certain segmentation techniques? Give 3 reasons.

1. For many applications of image processing, color information doesn't help us identify important edges or other features. There are ofcourse exceptions. This because of how the edge-detection algorithms work. 2. Complexity of the code. An application which has to check an image of multiple dimensions will be more complex to code than a 2 dimensional grayscale image. 3. Speed. With modern computers, and with parallel programming, it's possible to perform simple pixel-by-pixel processing of a megapixel image in milliseconds. Facial recognition, OCR, content-aware resizing, mean shift segmentation, and other tasks can take much longer than that.

What are the two Edge Detection Operators? Can you name some of the detection methods?

1. Gradient-based operator which computes first-order derivations in a digital image like, Sobel operator, Prewitt operator, Robert operator. 2. Gaussian-based operator which computes second-order derivations in a digital image like, Canny edge detector, Laplacian of Gaussian.

What tasks are included in image restoration? Name 3.

1. Image Processing. 2. Image Segmentation. 3. Image Enhancement.

How does the HVS work?

1. Light passes through the cornea and is bent/refracted to the lens. 2. When light passes thorough the lens, light is bent/refracted a second time. Double bending has reversed the image upside down. The image is then turned right side up in the brain. 3. Light finally hits the retina where an image is formed. In the Retina, we have Rods and Cones. These are the receptors that allows us to see. Rods are responsible for vision at low light levels (scotopic vision). They do not mediate color vision, and have a low spatial acuity. Cones are active at higher light levels (photopic vision), are capable of color vision and are responsible for high spatial acuity. The central fovea is populated exclusively by cones. When these cells detect light, they send signals to the brain. Cone cells help detect colors. There is also a blindspot in the back of the eye where there is no retina, this because the optical nerves that connect to the brain take this place. The image travels through the optic nerves as impulses to the brain.

What are the three ways of estimating the level of degredation in an image?

1. Observation 2. Experimentation 3. Mathematical modeling.

How many bands are usually used in Multispectral Imaging?

3 - 10

What is Gaussian Filter?

A Gaussian Filter is a low pass filter used for reducing noise (high frequency components) and blurring regions of an image. It is a realistic model of a defocused lens. The Gaussian Smoothing Operator performs a weighted average of surrounding pixels based on the Gaussian distribution. A couple other types of noise are: Salt-and-Pepper, Erlang (Gamma) Noise & Uniform Noise.

Describe different colourspaces

A color space can be described as a way of organizing colors. The reason for organizing the colors is to make it possible to use them in mediums such as on a computer or on a phone. This is done by mapping the real colors to the color-model's values. Some examples of color models are RGB and CMYK. A color space is therefore easier explained, grouped by the color model that is used for part of their specification. Adobe RGB is for an example a color space based on the RGB model. A lager color space allows for an image to contain a wider range of colors. The main and biggest reason we use different color spaces in computer vision is because there are different ways of representing colors. With the RGB color space, all pixels in an image have 3 different values. One value of each red, green, and blue colors. Now compare this with a grayscale image. A grayscale image is one in which each pixel contains a single sample representing the amount of light. For example, if you're tasked to find the brightest points in an image, it would be much easier to use the grayscale color space to find these points as each pixel represents a light intensity value.

What is the Midpoint Filter, and what is it used for?

A combination of the Min and Max filters. It works best for randomly distributed noise, like Gaussian or uniform noise

What is SIFT?

A computer vision algorithm to detect, describe, and match local features in images. Basic idea: - Take 16x16 square window around detected interest points (8x8 in example below) - Compute edge orientation (gradient angle minus 90°) for each pixel - Remove weak edges (below certain threshold at gradient magnitude) - Create histogram of surviving edge orientations (8 bins)

What is contours in image processing?

A contour is simply the boundary of an object in an image. Various representations of contours (e.g. chain code, Fourier descriptors, shape context) are used to recognize or categorize objects. Contours can a bit more complicated be explained simply as a curve joining all the continuous points (along the boundary), having same color or intensity.

What is an image feature?

A distinctive attribute or description of "something" we want to label or differentiate In the case for traffic, the "something" is vehicles.

What is a Feature Descriptor?

A feature descriptor is an algorithm which takes an image and outputs feature descriptors/feature vectors. Feature descriptors encode interesting information into a series of numbers and act as a sort of numerical "fingerprint" that can be used to differentiate one feature from another.

What is Object Detection?

A feature extration technique. Image/object detection places a bounding box around specific objects.

What is Image Segmentation?

A feature extration technique. In digital image processing and computer vision, image segmentation is the process of partitioning a digital image into multiple image segments, also known as image regions or image objects (sets of pixels).

What is Alpha-trimmed mean filter?

A hybrid of the Mean and Median filters. The basic idea behind filter is for any element of the signal (image) look at its neighborhood, discard the most atypical elements and calculate mean value using the rest of them. Alpha-trimmed mean filter is useful in situations where multiple types of noise appear, such as a combination of salt-and-pepper and Gaussian noise.

What is aperture?

A larger hole allows more light to hit the sensor, lightening your photos. A smaller hole allows less light to hit the sensor, darkening your photos. And by adjusting the aperture setting on your camera, you can adjust the size of the aperture (and, in turn, affect a photo's brightness). When light passes through a small hole, it does not travel in a straight line and is scattered in many directions (diffraction). Aperture also affects the depth of field.

Hva does a Mean filter do? Mention 3 types of Mean Filters

A mean filter is is a linear filter used to reduce noise by blurring, which you also can look upon as smoothing the image. With the method, each pixel value in an image is replaced with the average of all neighbouring pixels. It is performed using kernel convolution and a kernel that looks like: 1 1 1 1 1 1 1 1 1 The kernel is ran across the whole image, and each pixel-value in the kernel is multiplied by the value of the pixel in the kernel. All values in the kernel which now has been multiplied, is the divided by the size of the kernel, 1/9 usually. This resulting number will give you the average of neighbouring pixels. It is important that the new values are added to a new image, and not changing the current image. There are multiple versions of this filter: Arithmetic Mean Filter - The standard version. Well suited for random noise like Gaussian and uniform. Geometric Mean Filter - Achieves smoothing comparable to an arithmetic mean filter, but it tends to lose less image detail in the process. Harmonic Mean Filter - Works well for salt noise, but fails for pepper noise. Also works well with Gaussian noise.

What is Neighbourhood in image processing?

A pixel's neighborhood is some set of pixels, defined by their locations relative to that pixel, which is called the center pixel. Nightbourhood operations are commonly used in feature detection (line, edge, corner etc)

What is Adaptive filter?

An adaptive filter is a digital filter that has self-adjusting characteristics. It is capable of adjusting its filter coefficients automatically to adapt the input signal via an adaptive algorithm.

Explain the Sobel Operator

An edge detection mehod. Simply a case of trying to find the regions in an image where we have a sharp change in intensity or color. A high value indicates a steep change. Uses kernel convolution with a kernel that looks like this in x-direction: -1 0 1 -2 0 2 -1 0 1 You then have to do the same with a kernal in the y-direction that looks like this: -1 -2 -1 0 0 0 1 2 1

What is gradient in Computer Vision?

An image gradient is a directional change in the intensity or color in an image. The gradient of the image is one of the fundamental building blocks in image processing. For example, the Canny edge detector uses image gradient for edge detection.

What is ISO?

Artificial light added by the camera.

What is a grayscale image in computer vision?

As the name implies, the image contains gray-level information and no color information. The typical grayscale image contains 8-bit/pixel (bpp) data. In other words, this type of image has 256 different shades of colors in it, varying from 0 - 255, where 0 is black and 255 is white.

What is Refraction?

Bending of light. Light refracts whenever it travels at an angle into a substance with a different refractive index (optical density). This change of direction is caused by a change in speed. For example, when light travels from air into water, it slows down, causing it to continue to travel at a different angle or direction.

Explain Hough Transform.

Can detect lines, circles, ellipses and parabolas. The Hough transform takes a binary edge map as input and attempts to locate edges placed as straight lines. The idea of the Hough transform is, that every edge point in the edge map is transformed to all possible lines that could pass through that point. We assume that we have already gone through the steps of 1. Convert the image to grayscale. 2. Reduce noise - as the edge detection that using derivatives is sensitive to noise, we reduce it by blurring. 3. Calculate the gradient - helps identify the edge intensity and direction. and used canny edge detection. After that we map every point from the edge detection to Hough Space. Hough Space is this 2D space where every point is representing a line. A poll is then performed in this space where the lines that cross most of the other lines are defined as a line.

Explain Canny Edge detection.

Canny Edge Detection essentialy uses the Sobel Operator (another type of edge detection) and makes it better. It gets rid of the edges we're not interested in. The basic idea is that we need to look for a neighbourhood with strong signs of change. If a pixel has the intensity of 15 and its neighbouring pixel has an intensity of 80, there is a strong sign of change. And we have discovered a possible edge. To look at this visually it's most commonlyvisualized in matrices, where you can more easily see each pixel and its intensity. It is a multi-stage algorithm used to detect/identify a wide range of edges. The three first steps are performed på the sobel operator, the next steps it the canny edge detector: 1. Convert the image to grayscale. 2. Reduce noise - as the edge detection that using derivatives is sensitive to noise, we reduce it by blurring. 3. Calculate the gradient - helps identify the edge intensity and direction. 4. Non-maximum suppression - to thin the edges of the image: Uses local maximum: Finds out if a pixel is bigger than it's neighbours. 5. Hysterisis thresholding/Two level threshold - to identify the strong, weak and irrelevant pixels in the images. If the intensity of a value is under a certain strength, it is discarded. 6. Hysteresis edge tracking - helps convert the weak pixels into strong ones only if they have a strong pixel around them. Any pixel that is connected to any pixel above our top threshold is included by pixel traversal. Everything else we scrap. Extra: Since edge detection is susceptible to noise in the image, the first step is usually to remove the noise in the image with a filter like Median filter. thresholding

What is the difference between Classification, Detection and Segmentation?

Classification allows for the classification of a given image as belonging to one of a set of predefined categories. Object detection is a computer vision technique that allows us to identify and locate objects in an image or video. With this kind of identification and localization, object detection can be used to count objects in a scene and determine and track their precise locations, all while accurately labeling them. The principal goal of the segmentation process is to partition an image into regions (also called classes or subsets) that are homogeneous with respect to one or more characteristics or features.

What is Closing in terms of Morhpolical Operations?

Closing is dilation followed by erosion. Closing fills in small holes and gaps between connected components.

What is Color Image Processing?

Color image processing is the analysis, transformation, and interpretation of visual data presented in color. It can produce a range of results from a grayscale conversion of a black and white picture to a detailed analysis of information contained in a photograph taken by a telescope.

Can the computer match human perception?

Computers can be better at easy things; humans are much better at hard things.

What is contrast stretching?

Contrast stretching (often called normalization) is a simple image enhancement technique that attempts to improve the contrast in an image by `stretching' the range of intensity values it contains to span a desired range of values, e.g. the the full range of pixel values that the image type concerned allows.

How does Convolution work?

Convolution works by determining the value of the central pixel, by summing up the weighted values of all of the neighbor points together, and then creates a modified filtered image.

What is image degredation?

Degradation comes in many forms such as motion blur, noise, and camera misfocus. In cases like motion blur, it is possible to come up with a very good estimate of the actual blurring function and "undo" the blur to restore the original image.

How are digital images stored?

Digital images are stored as arrays of numbers. Images of RGB color spaces are also stored as arrays of numbers, but of multiple dimensions. These are Red, Green and Blue.

What are the two basic morphological operations?

Dilation and Erosion. Usinng a structural element over an object(sub-set of pixel of the image).

How does edge detection work?

Edge detection is a technique of image processing used to identify points in a digital image with discontinuities, simply to say, sharp changes in the image brightness. These points where the image brightness varies sharply are called the edges (or boundaries) of the image.

Name five different types of Image Segmentation Techniques

Edge-Based Segmentation (aka edge detection) Thresholding Segmentation Hough Transform Region-Based Segmentation (using multiple thresholds) Clustering-Based Segmentation

Explain the morphological operation Erosion (Nedsliting)

Erosion removes pixels on object boundaries. In other words, it shrinks the foreground objects. Enlarge foreground holes.

What is Feature Extraction?

Features are parts or patterns of an object in an image that help to identify it. For example — a square has 4 corners and 4 edges, they can be called features of the square, and they help us humans identify it's a square. Features include properties like corners, edges, regions of interest points, ridges, Feature Extraction = Feature Detection + description

What is Image Filtering?

Filtering is a neighborhood operation, in which the value of any given pixel in the output image is determined by applying some algorithm to the values of the pixels in the neighborhood of the corresponding input pixel. A pixel's neighborhood is some set of pixels, defined by their locations relative to that pixel.

What is the difference between Gaussian and Median Filters?

Gaussian filter is a linear type of filter which is based on Gaussian function. But the median filter is a non-linear type of filter. It preserves edges while removing noise.

What are Geometrical Transformations?

Geometric transformations are used to modify the spatial arrangement of pixels in an image

Explain the graph cut segmentation technique.

Graph cut is a semiautomatic segmentation technique that you can use to segment an image into foreground and background elements. Graph cut segmentation does not require good initialization. You draw lines on the image, called scribbles, to identify what you want in the foreground and what you want in the background. 1. The image is represented as a graph where each pixel represents a node. 2. Each node is connected to its neighbors as well as two virtual nodes - Foreground (F) and Background (B). 3. Weights are assigned to all graph connections. 4. Objective: find the minimum cut. - A cut is a set of edges that when removed separates F from B. GrabCut is an extension of Graph cut which used Iterated Graph Cuts.

What is High-Level processing?

High-Level Computer vision (image understanding) is a discipline that studies how to reconstruct, interpret and understand a 3D scene from its 2D images in terms of the properties of the structures present in the scene.

What is histogram equalization?

Histogram Equalization is a computer image processing technique used to improve contrast in images . It accomplishes this by effectively spreading out the most frequent intensity values, i.e. stretching out the intensity range of the image.

What is the difference between contrast stretching and histogram equalization?

Histogram equalization is about modifying the intensity values of all the pixels in the image such that the histogram is "flattened" (in reality, the histogram can't be exactly flattened, there would be some peaks and some valleys, but that's a practical problem). However, once histogram equalization is performed, there is no way of getting back the original image. In contrast stretching, there exists a one-to-one relationship of the intensity values between the source image and the target image i.e., the original image can be restored from the contrast-stretched image.

What are Histograms useful for?

Histograms help detect image acquisition issues • Over and under exposure • Brightness • Contrast • Dynamic Range Threshold selection Image Enhancement • Histogram equalization • Histogram stretching • Histogram matching

How many bands are usually used in Hyperspectral Imaging?

Hyperspectral imagery consists of much narrower bands (10-20 nm). A hyperspectral image could have hundreds or thousands of bands. In general, they don't have descriptive channel names.

What is Image Acquisition?

Image Acquisition is the first step in any image processing system. The general aim of any image acquisition is to transform an optical image (real-world data) into an array of numerical data which could be later manipulated on a computer. Image acquisition is achieved by cameras. We use different cameras for different applications. For normal images (family pictures, etc.), we use cameras that are sensitive to the visual spectrum.

What is the difference between Image Enhancement and Image Restoration?

Image Enhancement: - A process which aims to improve bad images so they will "look" better. Image Restoration: - A process which aims to invert known degradation operations applied to images.

What are the 6 steps of the computer vision pipeline?

Image acquisition -> pre processing -> feature extraction -> detection and segmentation -> high level processing - > application

What is Image Classification?

Image classification is the process of predicting a specific class, or label, for something that is defined by a set of data points. Image classification is a subset of the classification problem, where an entire image is assigned a label. Perhaps a picture will be classified as a daytime or nighttime shot. Probably one of the most well-known tasks in computer vision. It allows for the classification of a given image as belonging to one of a set of predefined categories. Let's take a simple binary example: we want to categorize images according to whether they contain a tourist attraction or not.

What is Image Compression?

Image compression is the application of data compression on digital images. In effect, the objective is to reduce redundancy of the image data in order to be able to store or transmit data in an efficient form. Image compression can be lossy or lossless.

What is Image Enhancement?

Image enhancement is the procedure of improving the quality and information content of original data before processing. Some Image Enchantment techniques are: - Filtering with morphological operations. - Contrast adjustment: Histogram equalization, Contrast Stretching (also called normalization). - Noise reduction: Gaussian Filter, Median Filter, Max and Min Filters. - Decrease blurring. - Smoothing and Sharpening filters (Mean filter, )

What is Image matching?

Image matching techniques are the techniques used to find existence of a pattern within a source image. Feature descriptors/feature vectors are compared. 1. Define a distance / similarity function to compare the two descriptors 2. Test against all features found in 𝐼2, get the one with min distance / max similarity. After a certain threshold 𝑡.

What is Image Restoration?

Image restoration is the operation of taking a corrupt/noisy image and estimating the clean, original image. Corruption may come in many forms such as motion blur, noise and camera mis-focus.

What is Billinear interpolation?

In computer vision and image processing, bilinear interpolation is used to resample images and textures. An algorithm is used to map a screen pixel location to a corresponding point on the texture map. A weighted average of the attributes (color, transparency, etc.)

What is Thresholding?

In digital image processing, thresholding is the simplest method of segmenting images. From a grayscale image, thresholding can be used to create binary images. The simplest thresholding methods replace each pixel in an image with a black pixel if the image intensity is less than a fixed value called the threshold, or a white pixel if the pixel intensity is greater than that threshold. In the example image on the right, this results in the dark tree becoming completely black, and the bright snow becoming completely white.

What is edge-based segmentation?

In edge-based segmentation, an edge filter is applied to the image, pixels are classified as edge or non-edge depending on the filter output, and pixels which are not separated by an edge are allocated to the same category. Edge-based segmentation and edge detection can be looked upon as the same. There are then subcategories of edge-based segmentation/edge-detection with all the different edge-detection techniques. Some of these techniques are: Canny Edge Detection, Sobel Edge Detection and Kirsh Edge Detection.

What is Dispersion?

In optics, and by analogy other branches of physics dealing with wave propagation, dispersion is the phenomenon in which the phase velocity of a wave depends on its frequency;[1] sometimes the term chromatic dispersion is used for specificity to optics in particular. A dispersive prism causes different colors to refract at different angles, splitting white light.

What is cluster based segmentation?

In this type of segmentation, we try to cluster the pixels with the same intensity levels/colors. The most common clustering technique is K-means.

What is image Interpolation?

Interpolation is the process of transferring image from one resolution to another without losing image quality. In Image processing field, image interpolation is very important function for doing zooming, enhancement of image, resizing any many more. Bicubic Interpolation:

What is the Inverse Filter?

Inverse Filtering is the process of receiving the input of a system from its output. It is the simplest approach to restore the original image once the degradation function is known. Easier explained: If we have a degraded image which is blurred, and we assume/guess a similar blurring filter which could have created this kind of blur, we can invert/remove the blur, by using algorithm in reverse for the blurring filter. Inverse filter is also a linear filter.

What is Kernal Convolution?

It is a process where we take a small grid of numbers and run it over the whole image, and transform the image based on what the numbers in the grid are. Different numbers in the kernel can give different results like blur, edge detection, sharpening, unsharpning etc The core of gaussian blur, mean blur and edge detection. En 3x3 Gaussian kernel ville hatt: 1 2 1 2 4 2 1 2 1 The further away you get from the pixel of interest, the less weight you have in the combined average.

What is Region-Based Segmentation?

It is the process of dividing an image into different regions based on the characteristics of pixels to identify objects or boundaries to simplify an image and more efficiently analyze it. Region growing is a simple region-based image segmentation method.

How does K-Means clustering work?

K-Means clustering algorithm is an unsupervised algorithm and it is used to segment areas of an image. It clusters, or partitions the given data into K-clusters or parts based on the K-centroids. The K-means clustering algorithm mainly performs two tasks: Determines the best value for K center points or centroids by an iterative process. Assigns each data point to its closest k-center. Those data points which are near to the particular k-center, create a cluster. The tighter the clusters are, the better the clustering. A byproduct of the clustering is a color palette of the image. K-means with 5 initial clusters, will segment the image into 5 segments/colors. K-means with 10 initial clusters will give a result with more colors. K-means step by step: 1. Choosing the number of clusters The first step is to define the K number of clusters in which we will group the data. Let's select K=3. 2. Initializing centroids Centroid is the center of a cluster but initially, the exact center of data points will be unknown so, we select random data points and define them as centroids for each cluster. We will initialize 3 centroids in the dataset. 3. Assign data points to the nearest cluster Now that centroids are initialized, the next step is to assign data points Xn to their closest cluster centroid Ck 4. Re-initialize centroids Next, we will re-initialize the centroids by calculating the average of all data points of that cluster. 5. Repeat steps 3 and 4 We will keep repeating steps 3 and 4 until we have optimal centroids and the assignments of data points to correct clusters are not changing anymore.

What is Linear filters?

Linear Filtering occurs when the operation performed on each pixel is a simple mathematical operation with a scalar where the result is similar for all pixels. Median filter is non-linear, gaussian filter is linear.

What is Low Level processing?

Low-Level processing operation involves tasks such as image preprocessing to reduce noise, contrast enhancement, image sharpening, etc. In the low-level process, both input and output are images. Blurring, contrast changing, segmentation and color changes are filters used in low level processing.

What are Morphological filter Min and Max Filters used for?

Min filters are useful for finding the darkest points in an image or for eroding light regions adjacent to dark areas. Reduces salt noise. Max filters are useful for finding the brightest points in an image or for eroding dark regions adjacent to bright areas. Reduces pepper noise. Say you have an binary image, and you run across every pixel with a 3x3 matrix. Min filter: The transformation replaces the central pixel with the darkest one in the running window. For example, if you have text that is lightly printed, the minimum filter makes letters thicker. Imagine a 3x3 array running accros an image, where the center pixel is replaced with the color of the neighbour pixel with the darkest color. Max filter: Whereas the minimum filter replaces the central pixel with the darkest one in the running window, the maximum filter replaces it with the lightest one. For example, if you have a text string drawn with a thick pen, you can make the sign skinnier.

What is nearest neightbour interpolation?

Nearest neighbour interpolation is the simplest approach to interpolation. Rather than calculate an average value by some weighting criteria or generate an intermediate value based on complicated rules, this method simply determines the "nearest" neighbouring pixel, and assumes the intensity value of it.

Is the Wiener filter a non-linear filter?

No. It's a linear filter.

What is Opening in terms of Morphological Operations?

Opening is erosion followed by dilation. For binary images, opening removes small connected components and small protrusions.

What is padding?

Padding describes the addition of empty pixels around the edges of an image. The purpose of padding is to preserve the original size of an image when applying a convolutional filter and enable the filter to perform full convolutions on the edge pixels. Types of padding: 1. Zero padding 2. Symmetrical padding 3. Circular padding

Name the three most common Image Acqusition systems

RGB, Spectral and 3D

What is Region growing?

Random pixels are selected in an image and are merged to neighbors if they are similar/homogeneous(same kind). Homogeneity of regions is used as the main segmentation criterion. The criteria for homogeneity: intensity, color, shape. Region growing is kinda like using the magic wand tool in photoshop to get an area of similar colors.

What two other types of noise is mentioned in class?

Rayleigh Noise and Erland (Gamme Noise)

What is Salt-and-Pepper Noise? What filter can be used to eliminate the black and white speckles?

Salt-and-pepper noise, also known as impulse noise, is a form of noise sometimes seen on digital images. This noise can be caused by sharp and sudden disturbances in the image signal. It presents itself as sparsely occurring white and black pixels. Use the Median Filter block to eliminate the black and white speckles in the image. Use the default parameters. The Median Filter block replaces the central value of the 3-by-3 neighborhood with the median value of the neighborhood. This process removes the noise in the image.

Sampling

Sampling is how often u get info from an image. how many pixels you have, while quantization is how you can represent the pixel, what they can be. The sampling rate determines the spatial resolution of the digitized image, while the quantization level determines the number of grey levels in the digitized image.

What is shutter speed?

Shutter speed is exactly what it sounds like: It's the speed at which the shutter of the camera closes. A fast shutter speed creates a shorter exposure — the amount of light the camera takes in — and a slow shutter speed gives the photographer a longer exposure.

What is Feature invariance?

Similarities between features in an image.

What is the most common technique/algorithm used to get a result of Superpixels? How does it work?

Simple Linear Iterative Clustering (SLIC) algorithm. 1. Makes random superpixels by sampling from a regular grid. Locations need to be in the center of the grids, not the edges. 2. For each cluster center, computes the distance to every pixel in the neighbourhood. 3. Assigned each pixel to the cluster that improves its current distance value. 4. Updates cluster centers. 5. Repeat 2,3,4.

What are Spectral Imaging?

Spectral imaging refers to a group of analytical techniques that collect spectroscopic information and imaging information at the same time. It uses multiple bands across the electromagnetic spectrum. While an ordinary camera captures light across three wavelength bands in the visible spectrum, red, green, and blue (RGB), spectral imaging encompasses a wide variety of techniques that go beyond RGB. Spectral imaging may use the infrared, the visible spectrum, the ultraviolet, x-rays, or some combination of the above. It may include the acquisition of image data in visible and non-visible bands simultaneously, illumination from outside the visible range, or the use of optical filters to capture a specific spectral range. It is also possible to capture hundreds of wavelength bands for each pixel in an image.

What are Superpixels?

Superpixels are the result of perceptual grouping of pixels, or seen the other way around, the results of an image segmentation. Segmentering av sammenhengende piksler med lignende farger.

Gradient calculation is a step in the canny edge algorithm. What does this do?

The Gradient calculation step detects the edge intensity and direction by calculating the gradient of the image using edge detection operators. Edges correspond to a change of pixels' intensity. To detect it, the easiest way is to apply filters that highlight this intensity change in both directions: horizontal (x) and vertical (y).

What edge-segmentation/edge-detection method is considered the best, and why?

The Sobel edge detector and Prewitt edge detector are able to detect edges but the edges detected are very less as compared to Canny edge detector. After all these results and comparative images, it is found that the performance of Canny edge detector is better than Sobel and Prewitt edge detector.

What is Wiener Filter?

The Wiener filter is the MSE-optimal stationary linear filter for images degraded by additive noise and blurring. Calculation of the Wiener filter requires the assumption that the signal and noise processes are second-order stationary (in the random process sense). It is used in Image Restoration and is the most important technique for removal of blur in images due to linear motion or unfocussed optics. In closing, it should be noted that Weiner filters are far and away the most common deblurring technique used because it mathematically returns the best results

What is Aliasing?

The aliasing effect is the appearance of jagged edges/sharp edges in an image rendered using pixels. Anti-aliasing is the smoothing of jagged edges in digital images by averaging the colors of the pixels at a boundary. Aliasing therefore un-smoothens/sharpens the image. Some methods of AA are: 1. Supersample Antialiasing 2. Multisampling Antialiasing

Explain the segmentation method called active contour.

The idea of Active Contours is to match a deformable curve to an image. External image forces pull the curve to the object contours, while the internal curve forces resist deformation. The algorithm explained:

What is illumination?

The illumination of the subject of a drawing or painting is a key element in creating an artistic piece, and the interplay of light and shadow is a valuable method in the artist's toolbox. The placement of the light sources can make a considerable difference in the type of message that is being presented.

What is Median Filter?

The median filter is the filtering technique used for noise removal from images and signals. Median filter is very crucial in the image processing field as it is well known for the preservation of edges during noise removal. The principle of the median filter is to replace the gray level of each pixel by the median of the gray levels in a neighborhood of the pixels, instead of using the average operation. For median filtering, we specify the kernel size, list the pixel values, covered by the kernel, and determine the median level. The Median Filter is considered the best-known "Order-Statistic Filter", which means the response of the filters is based on ranking values of the pixels in the neighbourhood encompassed by the filter.

Explain Probabilistic Hough Transform.

The method works by taking a sample of random points from an edge and running this group of points through a voting procedure. Instead of running a poll on all the points on the whole edge, it only takes a random set of points which obviously should reduce the workload. This works because a random subset of the total amount of points on the edge fairly accurate will represent the edge and surrounding noise and distortion. Probabilistic Hough Transform provides more flexibility and the possibility to detect edges more accurately if we just adjust the expected parameters in the code well enough.

Which spectrum is the visible spectrum a part of?

The visible spectrum is the segment of the electromagnetic spectrum that is visible to the human eye. Electromagnetic radiation in this range of wavelengths is called visible light or simply light. A typical human eye will respond to wavelengths from about 380 to about 750 nanometers.

Explain the segmentation method called Watershed.

This algorithm generally helps in detecting touching and overlapping objects/blobs in image. The algorithm explained:

What is a high-pass filter (sharpening)?

This is a technique used for image enhancement. A high-pass filter can be used to make an image appear sharper. A high-pass filter retains the high frequency information within an image while reducing the low frequency information. The kernel of the high pass filter is designed to increase the brightness of the center pixel relative to neighboring pixels. If there is no change in intensity, nothing happens. But if one pixel is brighter than its immediate neighbors, it gets boosted.

What is a low-pass filter (smoothing/blurring)?

This is a technique used for image enhancement. A low-pass filter, also called a "smoothing" filter, averages out rapid changes in intensity. The simplest low-pass filter just calculates the average of a pixel based on all of its immediate neighbors. The result replaces the original value of the pixel. The process is repeated for every pixel in the image. But why would you want a blurrier image? Often images can be noisy - no matter how good the camera is, it always adds an amount of "snow" into the image. Therefore a low-pass filter can sometimes be used to bring out faint details that were smothered by noise.

Explain the morphological operation Dilation (Utvidelse)

Used to enhance the features of an image. Dilation as a function requires two inputs, an image to be dilated, and a two dimensional structuring element. Dilation has many applications, but is most commonly used to exaggerate features in an image that would otherwise be missed. https://www.youtube.com/watch?v=7-FZBgrW4RE&ab_channel=PoonamKshirsagar

What is a binary image in computer vision?

What is a Binary Image? Binary images are images whose pixels have only two possible intensity values. Numerically, the two values are often 0 for black, and either 1 or 255 for white.

What is occlusion?

What is occlusion in computer vision? Occlusion often occurs when two or more objects come too close and seemingly merge or combine with each other. Image processing system with object tracking often wrongly track the occluded objects.

Is the Gaussian filter a linear filter?

Yes.

Is the Mean filter a linear filter?

Yes.