Scribd
Upload
11 views

Digital Image Classification

Remote Sensing

Uploaded by

Abha
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
11 views

Digital Image Classification

Remote Sensing

Uploaded by

Abha
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 33

DIGITAL IMAGE

CLASSIFICATION
What is Digital Image Classification
 Grouping of similar pixels
 Separation of dissimilar
ones
 Assigning class label to
pixels
 Resulting in manageable
size of classes

CLASSIFICATION METHODS
MANUAL
COMPUTER ASSISTED
STRATIFIED
Image classification

It is a science or technology which changes or turns the remote sensing data into
meaning categories.
Image classification is the process of assigning the land use land cover classes
with the help of pixel values of satellite imageries.
Why use it?
 To translate continuous variability of image data into map patterns that provide
meaning to the user.

 To obtain insight in the data with respect to ground cover and surface
characteristics.

 To find anomalous patterns in the image data set.


 Cost efficient in the analyses of large data sets

 Results can be reproduced

 More objective then visual interpretation

 Effective analysis of complex multi-band (spectral) interrelationships


Digital Image Classification
 Multispectral classification is the process of sorting pixels into a finite number of
individual classes, or categories of data, based on their data file values. If a pixel
satisfies a certain set of criteria , the pixel is assigned to the class that
corresponds to that criteria.
 Multispectral classification may be performed using a variety of algorithms
 Hard classification using supervised or unsupervised approaches.
 Classification using fuzzy logic, and/or
 Hybrid approaches often involving use of ancillary information.
Based on DN On the basis
value of geometry
SUPERVISED CLASSIFICATION
 The identity and location of some of the land cover types such as urban, agriculture, wetland are known a
prior through a combination of field work and experience. In supervised classification, you select some
representative samples for each land cover and land use class. The software then uses these signature values
and applies them to the entire image.

 The analyst attempts to locate specific sites in the remotely sensed data that represent homogenous
examples of these known land cover types known as training sites.

 Every pixel both inside and outside the training sites is evaluated and assigned to the class of which it has
the highest likelihood of being a member.

 We use some steps for this classification.

 Selection of area
 Generate signature file
 Classify
Supervised classification approach
SUPERVISED CLASSIFICATION

 In supervised training, you rely on your own pattern recognition


skills and a prior knowledge of the data to help the system
determine the statistical criteria (signatures) for data
classification.

 To select reliable samples, you should know some information—


either spatial or spectral—about the pixels that you want to
classify.
Training Samples and Feature Space Objects
 Training samples (also called samples) are sets of pixels that represent what is
recognized as a discernible pattern, or potential class. The system calculates
statistics from the sample pixels to create a parametric signature for the class.

Selecting Training Samples


 Training data for a class should be collected from homogeneous environment.

 if training data is being collected from n bands then >10n pixels of training data
is to be collected for each class.
Image space

Single-band Image Multi-band Image

• Image space (col,row)


• array of elements corresponding to reflected or emitted energy from IFOV
• spatial arrangement of the measurements of the reflected or emitted
energy
There are a number of ways to collect training site data-

 defining a polygon in the image

 using a class from a thematic raster layer from an image file of the
same area (i.e., the result of an unsupervised classification)
Evaluating Signatures
 There are tests to perform that can help determine whether the
signature data are a true representation of the pixels to be
classified for each class. You can evaluate signatures that were
created either from supervised or unsupervised training.
Supervised classification algorithm

Selecting appropriate classificationalgorithm


1. Maximum likelihood
2. Minimum distance
3. Parallelopiped
4. Principal component
5. Support vector machine(SVM)
6. ISO cluster
PARALLELEPIPED CLASSIFICATION
ALGORITHM
 In the parallelepiped decision rule, the data file values of the candidate
pixel are compared to upper and lower limits. These limits can be
either:
1. the minimum and maximum data file values of each band in the
signature,
2. the mean of each band, plus and minus a number of standard
deviations, or
3. any limits that you specify, based on your knowledge of the data and
signatures.
 There are high and low limits for every signature in every band. When
a pixel’s data file values are between the limits for every band in a
signature, then the pixel is assigned to that signature’s class.
Overlap Region
In cases where a pixel may fall into the overlap region of two or more
parallelepipeds, you must define how the pixel can be classified.

 The pixel can be classified by the order of the signatures.

 The pixel can be classified by the defined parametric decision rule.

 The pixel can be left unclassified.


MINIMUM DISTANCE TO MEANS
CLASSIFICATION ALGORITHM
 This decision rule is computationally simple and commonly used.
 Requires mean vectors for each class in each band μck from the
training data.
 Euclidean distance is calculated for all the pixels with all the signature
means
D = √ (BVijk- μck)2 + (BVijl- μcl)2
Where
μck and μcl represent the mean vectors for class c measured in bands k
and l
 Any unknown pixel will definitely be assigned to one of any classes,
there will be no unclassified pixel.
Maximum Likelihood/Bayesian
Decision Rule
 The maximum likelihood decision rule is based on the probability that a pixel
belongs to a particular class. The basic equation assumes that these
probabilities are equal for all classes, and that the input bands have normal
distributions.

 If you have a priori knowledge that the probabilities are not equal for all
classes, you can specify weight factors for particular classes. This variation of
the maximum likelihood decision rule is known as the Bayesian decision rule
(Hord, 1982).
Advantages
 The most accurate of the classifiers (if the input samples/clusters have a
normal distribution), because it takes the most variables into
consideration.
 Takes the variability of classes into account by using the covariance
matrix, as does Mahalanobis distance.
Disadvantages
 An extensive equation that takes a long time to compute. The
computation time increases with the number of input bands.
 Maximum likelihood is parametric, meaning that it relies heavily on a
normal distribution of the data in each input band.
 Tends to overclassify signatures with relatively large values in the
covariance matrix.
Unsupervised Classification in Remote Sensing
In Unsupervised classification we give a command to the software according to
our objectives and purposes. First of all the software converts the group’s pixels
into clusters. Then the software classifies each cluster with a land-use and land
cover class.

The identities of land cover types to be specified as classes within a scene are
generally not known to prior because ground reference information is lacking or
surface features within the scene are not well defined.

The computer is required to group pixels with similar spectral characteristics into
unique clusters according to some statistically determined criteria.

Analyst then combine and relabels the spectral clusters


into information classes. We use two steps for unsupervised classification.
1. Selection of area
2. Assign classes
UNSUPERVISED CLASSIFICATION
 It requires only a minimum amount of initial input from the analyst.
 Numerical operations are performed that search for natural groupings of the
spectral properties of pixels.
 User allows computer to select the class means and covariance matrices to be
used in the classification.
 Once the data are classified, the analyst attempts a posteriori to assign these
natural or spectral classes to the information classes of interest.
 Some clusters may be meaningless because they represent mixed classes.
 Clustering algorithm used for the unsupervised classification generally vary
according to the efficiency with which the clustering takes place.
 Two commonly used methods are-
1.Chain method
2.Isodata clustering
CHAIN METHOD
 Operates in two pass mode( it passes through the registered multispectral
dataset two times).

 In the first pass the program reads through the dataset and sequentially builds
clusters.

 A mean vector is associated with each cluster.

 In the second pass a minimum distance to means classification algorithm is


applied to whole dataset on a pixel by pixel basis whereby each pixel is
assigned to one of the mean vectors created in pass 1.

 The first pass automatically creates the cluster signatures to be used by


supervised classifier.
PASS 1: CLUSTER BUILDING

 During the first pass the analyst is required to supply four types of
information-
 R , the radius distance in spectral space used to determine when a new cluster
should be formed.
 C, a spectral space distance parameter used when merging clusters when N is
reached.
 N , the number of pixels to be evaluated between each major merging of
clusters.
 Cmax maximum no. of clusters to be identified.

PASS 2: Assignment of pixels to one of the Cmax clusters using


minimum distance classification logic
Q1. Classification of remotely sensed data is
a. Information extraction process
b. Correcting errors in the image
c. Application of enhancement algorithms to the image

Q2. Range of transformed divergence is


a. 0-100.
b. 0-200
c. 0-2000

Q3. Which of following is supervised classification method


a. Isodata
b. Maximum Likelihood Classifier
c. Chain method
Method of object-based image analysis
We use different methods to classify images.

1. Shape
2. Texture
3. Spectral signature
4. Geographic context
5. Nearest neighbour classification
Question/Interaction
How can you differentiate between supervised and
unsupervised classification

When should one use unsupervised classification

You might also like