Noises net effect is a corrupted image that needs to be preprocessed to reduce or eliminate the noise. Systematic noises come from dirty lenses, faulty electronic components, and low resolution. Random noises caused by environmental effects or bad lighting.
Noise Reduction Operations
The noise is reduced by using masks. Create masks that behave like a lowpass filter, such that the higher frequencies of an image are attenuated while the lower frequencies are not changed very much.
A number of images of the exact same scene are averaged together. This technique is time consuming. This technique is not suitable for operations that are dynamic and change rapidly. It is more effective with an increased number of images. It is usefull for random noise.
When the Fourier transform of an image is calculated, the frequency
spectrum might show a clear frequency for the noise, which in any cases can be selectively eliminated by proper filtering.
In median filter the value of the pixel is replaced by the median of the values of the pixels in a mask around the given pixel, stored in ascending order. A median is the value such that half of the values
in the set are below and half are above the median. The median is stronger in eliminating spikelike noises without blurring the object or decreasing the overall sharpness of the image. The median is independent of the value of any single pixel in the set.
Class of routines and techniques that operate on an image and result in a line drawing of the image. That requires much less memory to be stored, much simpler to be processed, and saves in computation and storage costs.
The lines represent changes in values such as cross section of planes, intersections of planes,…. All techniques used operate on differences between the gray levels of pixels or group of pixels through masks or thresholds.
These are techniques of segmentation. Through these techniques an attempt is made to separate the different parts of an image into components with similar characteristics that can be used in further analysis. Segmentation by regions will result in complete and closed boundaries.
Regios Growing Techniques
Two approaches are used for region segmentation:
Region growing by similar attributes, such as grey-level ranges or other similarities.
Region splitting into smaller areas by using finer differences.
A collection of operations and techniques that are used to extract
information from images. Among these are feature extraction, object recognition, analysis of the position, size, orientation, and extraction of depth information.
In vision applications distinguishing one object from another is accomplished by means of features that uniquely characterize the object. A feature [area, diameter, perimeter], is a single parameter that permits ease of comparison and identification. An important objective in selecting these features is that the features should not depend on position or orientation.
Feature Extraction Techniques
The techniques available to extract feature values for twodimensional
cases can be roughly categorized as those that deal with boundary features and those that deal with area features.
The next step in image data processing is to identify the object the
image represents. This identification is accomplished using the extracted feature information described. The recognition algorithm must be powerful enough to uniquely identify the object.
Object recognition by Features
This may include gray-level histogram, morphological features such as area, perimeter, number of holes, eccentricity, cord length, moments,…. The information extracted is compares with a prior information about the object, which may be in a lookup table.
Basic Morphological Features Used For Object Identification
The average, maximum, or minimum gray levels.
The perimeter, area, diameter of an object, number of holes it has and other morphological characteristics.
The minimum aspect ratio (the ratio of the width to the length of a rectangle enclosed about the object). Thinness [(perimeter)^2/area or diameter/area] and Moments.