I am getting cofused by the basic definition of feature vector. It is too technical
in simple words how will you define feature vector?
A vector is a series of numbers. It is like a matrix with only one row but multiple columns (or only one column but multiple rows). An example is: [1,2,3,5,6,3,2,0].
A feature vector is just a vector that contains information describing an object's important characteristics.
In image processing, features can take many forms. A simple feature representation of an image is the raw intensity value of each pixel. However, more complicated feature representations are also possible. For facial expression analysis, I use mostly SIFT descriptor features (scale invariant feature transform). These features capture the prevalence of different line orientations.
Input vector i.e input image in (2D spatial representation) or the numerical value that form the image is "feature vector". It has or you will want to find "features" or "attributes" by identifying unique characteristics in the image. Technicality comes when you start looking for specific features or start with transformation to identity the features else it is just a collection of numbers (real or integers) !
A vector is a series of numbers. It is like a matrix with only one row but multiple columns (or only one column but multiple rows). An example is: [1,2,3,5,6,3,2,0].
A feature vector is just a vector that contains information describing an object's important characteristics.
In image processing, features can take many forms. A simple feature representation of an image is the raw intensity value of each pixel. However, more complicated feature representations are also possible. For facial expression analysis, I use mostly SIFT descriptor features (scale invariant feature transform). These features capture the prevalence of different line orientations.
In my experience a feature vector is a collection of features extracted from the image, for example in fingerprint recognition a feature vector is a collection of minutiae extracted from the fingerprint.
A feature vector is a one dimensional matrix which is used to describe a feature of an image. It can be used to describe an entire image (Global feature) or a feature present at in a location in the image space (local feature).
let us take simple example of color image, color of an object is also represented as one the feature. then it can be represented as f=[r,g,b] where r, g, b are corresponding values of pixel in three different planes which represents the color feature of the particular pixel. Likewise collection of other features that may be related with texture, object etc considered as feature vector.
hi,
feature vector is a collection of features which you extracted from the input data (image) in the form of matrix. We can categories those features in to global and local features.
We have several feature extraction algorithms please go through the following links
1.http://shonen.naun.org/multimedia/NAUN/bio/bio-2.pdf
2.http://nature.berkeley.edu/~penggong/PDFpapers/GongRSE92.pdf
3. http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6100285&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D6100285
4.http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=766860&punumber=83&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel5%2F83%2F16615%2F00766860.pdf%3Ftp%3D%26arnumber%3D766860%26punumber%3D83
5.http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=974061&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D974061.
thank you.
It usually does not mean anything more than a collection of features computed from the image. Often this is used for tasks like comparison, classification, etc instead of the original image. Features can be average density, centroid, number of black pixels in various quadrants, and so on....
This is a good question.
From what has already been written, the term "vector" needs to be clarified before we can begin to tackle the question of a feature vector.
The term vector was introduced by Saunders and MacLane in their 1950s algebra book. Before that, mathematicians referred to points in an n-dimensional linear space rather than vectors in an n-dimensional vector space. Either points or vectors are central to linear algebra. The more usual term is linear space rather than vector space. Engineers tend to favour the term "vector" and mathematicians tend to favour "point".
The coordinates of a point in an n-dimensional linear space identify a location in the space. In answer, to this question, we need to consider an n-dimensional feature space, where the coordinates of each point are usually real feature values. So, for example, if we want to work with objects that have 5 features, then we will want to work and compare points, each with 5 coordinates, in a 5-dimensional feature space.
The magical part of this stems from the choice of probe functions that define the features of objects in an object space. Each probe function is a mapping from an object such as a digital image pixel to a real value (feature value). With 5 functions, we can then begin considering feature vectors, each with 5 values. In all of this, it is important to stay close to the pulse of linear algebra and the mathematics of (finite) n-dimensional linear spaces. See, for example,
Paul R. Halmos, Finite Dimensional Vector Spaces, Princeton University Press, 1947:
http://press.princeton.edu/about_pup/PUP100/book/1jHalmos.pdf
At one time, it was possible to download a copy of this book. Does anyone have the URL for the downloadable copy?
You can say that a feature vector represents a particular object/class in a feature space. Building such a feature space like in SVM it transform (including weight vector, bias etc) the data into high dimension spaces which is actually called a feature space. In my understanding I would say there is a transfer function who put some logic on the Input Vector from sample space to transform it to feature vector.
@Rakesh Singh: please explain me with some images and videos ......
A good place to start is a greyscale digital image. Let Image denote such an image. Here are a list of features:
Intensity: Every pixel has an intensity image(x,y) at location (x,y), usually in the range from 0 to 255.
gradient direction and gradient magnitude:
[gradx, grady] = imgradientxy(image); % gradient magnitudes in x and y directions
[Gmag,Gdir] = ingredient(gradx, grady); % gradient direction for each pixel
That gives a feature vector (gradx, grady, Gmag, Gdir) for the pixels in the image.
Try
image = rgb2gray(imread('gray_image.bmp')); % Try the colour for Lena
[gmag, gdir] = imgradient(image);
% Display images
figure, imshow(image);
figure, imshowpair(gmag, gdir, 'montage');
figure, imshowpair(gradx, grady, 'montage');
from
http://www.mathworks.com/help/images/ref/imgradientxy.html#btp1osx-7
The gmag and gdir image for Lena is shown in the attached images.
A feature vector is just a vector containing multiple elements (features). The features may represent a pixel or a whole object in an image. Examples of features are color components, length, area, circularity, gradient magnitude, gradient direction, or simply the gray-level intensity value. It depends on which features are useful for the application at hand. Some people compute special features using image processing and computer vision techniques and some people just use the original pixel intensities as features.
Example: v = [R; G; B]; is a feature vector containing color components of a pixel or an object.
In a typical object recognition application, feature vector of a query object is compared with that of each object in a database to know how much the query object matches with each object in the database. There are many techniques to compare two feature vector. One of them is just the Euclidean distance between the feature vectors of two objects.
I hope it helps.
A feature vector is the one containing several different elements (features). The features may be associated to a pixel, a connected component or an object in an image. These characteristics can be viewed asthe functions that provide relevant informations. In most cases, these features are used in the treatments related to the classification and pattern recognition.
In the field of syntaxical image analysis, the examples of functions are usually colored components, length, area, compactness, amplitude and direction of the gradient.
in the case of segmentation some applications are simply interested in the pixel gray level value as a feature.
So the feature vector depends on what features are useful for the application. Some people calculate specific characteristics using image processing techniques and computer vision and some others simply use the intensities of the original pixels as features.
it is important to note that , any application implemented to solve a given problem needs the use of analytical tools, These latters require the handling features that may be intrinsic to the image or can be synthesized to establish a knowledge base for the processing algorithm.
All images stored matrix form, but if you extract all image or a section of images, you can convert this matrix to vector. (a matrix that have 1column and many row). In public, features of a image process and store in vector form, that we tell that: feature vector.
Features are measurable properties, distinguishing characteristic (in vectorial form) of the phenomena being observed. In the context of image processing it correspond to vectors representing color, texture, specral information, edge, different patterns, etc...
Feature vector is a collection of the important information that describe that object in the image , such as pixel intensities , mean, shape, edges , and so on..
lets more specific to image analysis, feature vector is the output of convolution of any filter with the input image to get some patterns/features of objects in image.
- Badges
- Science topic