Pulse-coupled neural networks (PCNN)

During the last few years there was a shift of the emphasis in the artificial neural network community toward spiking or pulse-coupled neural networks. Motivated by biological discoveries, many studies consider pulse coupled neural networks with spike-timing as an essential component in information processing by the brain.
           Pulse-coupled neural networks (PCNN) were introduced as a simple model for the cortical neurons in the visual area of the cat's brain.These neural models are proposed by Eckhorn and Johnson.
.            The essential model of PCNN, is described with details, that can be implemented to perform a number of  digital image processing applications.
            We describe in the next sections a model that evaluates the global pulse of a PCNN in order to find correlation in the pulse signal and achieve pattern recognition.

Image processing is any form of information processing for which the input is an image, such as photographs or frames of video; the output of image processing can be either an image or a set of characteristics or parameters related to the image. Most image-processing techniques involve treating the image as a two-dimensional signal and applying standard signal-processing techniques to it.
Basic purpose of image processing is for:
1.          Improvement of pictorial information for human interpretation.

2.          Processing of image data for storage, transmission, and representation for autonomous machine perception.

            A PCNN is a two-dimensional neural network. They are treated as the third generation of NN models, that takes in to account spiking nature of neurons. Each neuron in the processing layer is directly tied to an image pixel or a set of neighboring image pixels, the two linking and feeding inputs are iteratively processed and together to produce a pulse image with features, that can be changed by varying the PCNN parameters.
A Simple Neuron

A neuron is a device with many inputs and one output. The neuron has two modes of operation; the training mode and the using mode. In the training mode, the neuron can be trained to fire (or not), for particular input patterns. In the using mode, when a taught
input pattern is detected at the input, its associated output becomes the current
output. If the input pattern does not belong in the taught list of input patterns, the firing rule is used to determine whether to fire or not.
In PCNN,a neuron is operated in using mode.

   In the field of digital image processing  and pattern recognition , traditional models are
either subject to problems determined by geometric transforms (scaling, translation or rotation) or to high computational complexity.
      Moreover, it is known today that parallel processing could solve determined by geometric transforms to take advantage of it we need parallelisable models.
Neural models fits this requirement.

When PCNN is applied in image processing, it is a single layer two dimensional array of laterally linked neurons.


¨       The number of neurons in the network is equal to the number of input image. One-to-one correspondence exists between image pixels and neurons.

 ¨       Each pixel is connected to a unique neuron and each neuron is connected with the
surrounding neurons with a radius of linking field.
¨         The neuron receives input signals from other neurons and
from external sources through the receptive fields.
  ¨         After the receptive fields have collected the inputs, they are divided into two or more internal channels. One channel is the feeding input F and the other is the linking
input L.
¨           The feeding connections are required to have a slower characteristic response time constant than those of the linking inputs.
¨           The linking inputs are biased and then multiplied together, and further multiplied with the feeding input to form the total internal activity U.
¨        The pulse generator of the neuron consists of a stepfunction
generator and a threshold signal signal generator.
¨         At each time step the neuron output .
¨         Y  is set to 1 when the internal
activity U is greater than the threshold function T. The threshold input at each time step is updated.
¨          The output of the neuron is consequently reset to zero when T is larger than U. Thus at one time step the pulse generator produces a single pulse at its output whenever the value of U exceeds T.

  • Segment  Ability
Because of the local interconnections between the neurons, neurons encourage their neighbours to fire only when they fire.
Thus, if a group of neurons is close to firing then one neuron can trigger the entire group. Thus, similar segments of
the image fire in unison. This creates the segmenting ability of the PCNN.
·         Availability of Texture information
The edges have differing neighboring activity than do the interior of the object. Thus, the edges, which will still fire in unison, but will do so at different times than do the interior segments. Thus, the edges are may be isolated. After several iterations the groupings of neurons tend to break in time. This “break-up” is dependent upon the texture within a segment. This is caused by minor differences that eventually propagate (in time) to alter the neural potentials. Thus, texture information becomes

  • Denoising
For denoising, the intensity of a noisy pixel is significantly different from the
intensities of its surrounding pixels. Therefore, most neurons corresponding to noisy pixels do not capture neighboring neurons or get captured by the neighboring neurons.

No comments:

Post a Comment

leave your opinion