11.6. RECURSIVE FILTRATION

Lecture



In the previous sections of this chapter, transformation processing was considered as an indirect method for performing two-dimensional linear processing. It has been shown that conversion processing often requires much less arithmetic than with standard methods. In this section, another linear processing method called recursive filtering will be considered [14–16]. Sometimes recursive filtering is even more efficient than processing with conversion. In addition, in this case, data storage requires a smaller storage capacity than during processing with conversion.

Recursive filtering is based on a recurrent relationship between system input and output variables. For one-dimensional signals, the following recurrence relation has the following form [14]:

  11.6.  RECURSIVE FILTRATION , (11.6.1)

Where   11.6.  RECURSIVE FILTRATION ,   11.6.  RECURSIVE FILTRATION - samples of the input sequence,   11.6.  RECURSIVE FILTRATION ,   11.6.  RECURSIVE FILTRATION - samples of the output sequence, and   11.6.  RECURSIVE FILTRATION and   11.6.  RECURSIVE FILTRATION - weight factors. The key point here is that   11.6.  RECURSIVE FILTRATION th element of the output sequence depends not only on the last and   11.6.  RECURSIVE FILTRATION the penultimate elements of the input sequence, but also from   11.6.  RECURSIVE FILTRATION previous elements of the output sequence.

Most methods of synthesis and analysis of recursive filters are based on the use of   11.6.  RECURSIVE FILTRATION -conversion. By definition [17, 18]   11.6.  RECURSIVE FILTRATION -transformation   11.6.  RECURSIVE FILTRATION -element sequence   11.6.  RECURSIVE FILTRATION gives an image

  11.6.  RECURSIVE FILTRATION . (11.6.2)

It is not difficult to show that   11.6.  RECURSIVE FILTRATION -transformation of expression (11.6.1) gives the image

  11.6.  RECURSIVE FILTRATION , (11.6.3)

Where   11.6.  RECURSIVE FILTRATION and   11.6.  RECURSIVE FILTRATION - images of the corresponding sequences.

Two-dimensional recursive filtering is based on the following recurrent relationship between the input and output arrays [19, 20]:

  11.6.  RECURSIVE FILTRATION (11.6.4)

Where   11.6.  RECURSIVE FILTRATION - input array from   11.6.  RECURSIVE FILTRATION elements   11.6.  RECURSIVE FILTRATION - output array from   11.6.  RECURSIVE FILTRATION elements as well   11.6.  RECURSIVE FILTRATION and   11.6.  RECURSIVE FILTRATION - weight factors. It is assumed that the recursive filtering process starts from the upper left corner of the input array. Using two-dimensional   11.6.  RECURSIVE FILTRATION -transformations from equality (11.6.4) is obtained

  11.6.  RECURSIVE FILTRATION , (11.6.5)

Where   11.6.  RECURSIVE FILTRATION and   11.6.  RECURSIVE FILTRATION - two-dimensional images of the corresponding arrays. For example,

  11.6.  RECURSIVE FILTRATION . (11.6.6)

During the synthesis of recursive filters, it is required to choose such arrays of weighting factors.   11.6.  RECURSIVE FILTRATION and   11.6.  RECURSIVE FILTRATION to output array   11.6.  RECURSIVE FILTRATION turned out to be equivalent to the array resulting from the convolution of the function   11.6.  RECURSIVE FILTRATION describing the original image with a given impulse response   11.6.  RECURSIVE FILTRATION . As a rule, an exact coincidence of arrays cannot be achieved, and it is necessary to use approximate methods in the synthesis of filters [21, 22]. This raises the question of the stability of the calculated recursive filter. If the filter is unstable, then rounding errors or noise present in the input array may not be attenuated during processing, but may increase to a very large level. A recursive filter is stable [20] if the coefficients   11.6.  RECURSIVE FILTRATION expansion of the frequency response of the filter in a series in powers of the variables   11.6.  RECURSIVE FILTRATION and   11.6.  RECURSIVE FILTRATION

  11.6.  RECURSIVE FILTRATION (11.6.7)

are absolutely summable, i.e., they satisfy the condition

  11.6.  RECURSIVE FILTRATION . (11.6.8)

Several methods of testing recursive filters for stability have been developed [23–25]. When processing the image according to equality (11.6.4) is required

  11.6.  RECURSIVE FILTRATION (11.6.9)

arithmetic operations. Here   11.6.  RECURSIVE FILTRATION and   11.6.  RECURSIVE FILTRATION - sizes of arrays of weight factors for the input and output images, a   11.6.  RECURSIVE FILTRATION - dimensions of the input image. If the images and arrays of the weighting factors are square (i.e.   11.6.  RECURSIVE FILTRATION ,   11.6.  RECURSIVE FILTRATION and   11.6.  RECURSIVE FILTRATION ), then with recursive filtering you need to run

  11.6.  RECURSIVE FILTRATION (11.6.10)

operations. For comparison, we point out that to obtain the final convolution (see Section 9.3) it is required

  11.6.  RECURSIVE FILTRATION (11.6.11)

operations where   11.6.  RECURSIVE FILTRATION - the size of the impulse response. As shown in sect. 11.2, to obtain convolution using the fast Fourier transform, approximately

  11.6.  RECURSIVE FILTRATION (11.6.12)

operations. The degree of relative efficiency of the three types of processing depends on the size of the pulse response array [27]. If these dimensions are small, then a direct number of arithmetic operations is required for the direct method of obtaining convolution and recursive filtering. In this case, it is possible to compare the effectiveness of both methods with the efficiency of obtaining convolution using FFT using the graph in fig. 11.3.2. For large sizes of the impulse response array, the fast method of obtaining convolution turned out to be much more efficient than the direct method. Comparing the number of operations in equalities (11.6.10) and (11.6.12), it can be shown that with large impulse response sizes, recursive filtering turns out to be more efficient than the fast method of obtaining convolution if the areas of the arrays of the recursive filter satisfy the inequality

  11.6.  RECURSIVE FILTRATION , (11.6.13)

Where   11.6.  RECURSIVE FILTRATION - constant, taking values ​​from 1 to 20, and its value depends on the type of FFT algorithm used and on how much the calculations with complex numbers occur more slowly than with real ones.

created: 2016-09-09
updated: 2021-03-13
132385



Rating 9 of 10. count vote: 2
Are you satisfied?:



Comments


To leave a comment
If you have any suggestion, idea, thanks or comment, feel free to write. We really value feedback and are glad to hear your opinion.
To reply

Digital image processing

Terms: Digital image processing