Noise in image refers to the irregular distortion or variation in luminance or color of the image.
Occurrence of noise in image is usually unavoidable due to scanner, printer or digital camera error.
Therefore it may happen that the noise gets converted into black foreground spot during binarization and which could be mistaken for some input character.
Thus it is important to filter out the noise.
For this different types of filters depending upon the application are applied.
In this step we have used Median Filter to remove salt and pepper noise from the image in case of Car number plate recognition and handwritten character image recognition.