Digital image processing

The digital image processing is the set of techniques applied to digital images in order to improve quality or facilitate the search for information.
Filtering process
It is the set of techniques encompassed within the pre-processing images whose main objective is to obtain, from a source image, one end of which result is more appropriate for a specific application to improve certain features of it that makes it possible to carry out operations of processing on it .
The main objectives to be achieved by the application of filters are:
Soften the image: reduce the amount of intensity variations between neighboring pixels.
Remove Noise: remove those pixels whose intensity level is very different from its neighbors and whose origin can be both in the process of acquiring the image and the transmission.
Enhance edges: to highlight the edges that are located in an image.
Detect edges: detect the pixels where an abrupt change in the intensity function.
Therefore, the filters are considered as transactions that are applied to the pixels of a digital image to be optimized, emphasize certain information or a special effect in it.
The filtering process can be performed on the frequency domain and / or space.
Type


There are basically three different types of filters that can be applied:
Lowpass filter: attenuates high frequencies and low remains unchanged. The result in the spatial domain is equivalent to a smoothing filter, where high frequencies are filtered correspond to strong changes of intensity. Get reduce noise smoothing existing transitions.
Highpass filter: attenuates low frequencies while keeping unchanged the high frequencies. Since high frequencies correspond to the images to sudden changes in density, this type of filter is used because, among other advantages, offers improved edge detection in the spatial domain, as these contain lots of these frequencies. Reinforces the contrasts found in the image.
Bandpass filter: attenuates very high or very low frequencies while maintaining midrange band.
Advantage
Simple and easy to implement method.
Easy concept association with certain frequency characteristics of the image; soft key changes involving low frequencies and high frequencies swings.
It provides flexibility in designing solutions for this search.
Filtering speed when using the convolution theorem.
Disadvantages [edit]
Knowledge is required in various fields to develop an application for image processing.
The noise can not be completely eliminated.
Domain filtering space
Filtering operations are performed directly on the image pixels. In this process relates to each and every one of the points of the image, a set of nearby pixels to the target pixel in order to obtain useful information, dependent on the type of applied filter, which allows to act on the particular pixel in which it is carrying out the filtering process to thereby obtain improvements over the image and / or data that could be used in future actions or work processes on it.
The filters in the spatial domain can be classified into:
Linear filters (or masks based on convolution kernels).
Nonlinear filters.
The concept is understood as a kernel coefficient matrix where the environment of the point (x, y) is considered in the image for g (x, y) is determined by the size and shape of the selected kernel. Although the shape and size of the matrix is ​​variable and is the choice of each user is common to use nxn square kernels. Depending on the implementation, in the image boundaries special treatment is applied (an outer frame or assumed zero values ​​of the edge are repeated) is applied or none. It is for this reason that the type of filtering is set by the content of the kernel used.