JP4982393B2 - Image filtering apparatus, image filtering program, image filtering method, and ultrasonic diagnostic apparatus - Google Patents

Description

The present invention relates to an image filtering apparatus, an image filtering program, and an image filtering method, and more particularly to an image filtering apparatus , an image filtering program , and an image filtering in an ultrasonic diagnostic apparatus that performs imaging based on an ultrasonic Doppler signal obtained from a subject. The present invention relates to a method and an ultrasonic diagnostic apparatus .

  Of the filter processing in raster images, the median filter (median extraction filter) is known to perform noise removal, pixel loss reduction, and smoothing while maintaining the boundary between resolution and image shape. ing. As an application example of the median filter, there is a color Doppler method in an ultrasonic diagnostic method.

  The color Doppler method scans a predetermined section in a living body with ultrasonic pulses, and when the ultrasonic wave is irradiated to a reflector that moves blood (blood cells) or the like, the velocity of the reflector (blood flow velocity) The Doppler frequency shift generated in response to the above is captured and imaged. In order to improve the diagnostic ability in the above-mentioned color Doppler method, excellent measurement accuracy (especially low flow rate detection ability), time resolution, and spatial resolution are required, but these requirements are simultaneously highly satisfied. It is difficult to make it. Therefore, for spatial resolution degradation, the two-dimensional median filter (median extraction filter) smoothes the boundaries of blood vessels and blood flow distributions, removes noise, and causes image omissions caused by ultrasonic interference and the like. (So-called blackout pattern) is reduced (see, for example, Patent Document 1 below).

  However, the median processing is practically data sorting (rearrangement), and requires a large number of calculations for comparison and replacement. Therefore, various methods have been devised to reduce the number of sort calculations for median processing.

For example, with respect to data of 3 rows and 3 columns, a method is known in which each column is first sorted, then each row is sorted, and finally three pieces of diagonal data are sorted to obtain their median values ( For example, see Patent Document 2 below). A method of sorting data up to a maximum of 5 rows and 5 columns other than 3 rows and 3 columns is shown in Non-Patent Document 1 below.
JP 2000-262520 A JP-A-11-149554 “Fast Algorithms for Median Filtering”, Intel Corporation, 2001.

  The algorithm described in Patent Document 2 and “Fast Algorithms for Median Filtering”, Intel Corporation, 2001. is used when simultaneously calculating a plurality of data using a SIMD (Single Instruction Multiple Data) processor or the like. It is effective. However, it still requires a large number of calculations for median processing.

Accordingly, the present invention has been made in view of the above circumstances, and its purpose is to achieve smoothing and noise removal image drop reduction similar to those of a median filter, and to perform high-speed image filtering with fewer calculation times than the median filter. An apparatus , an image filtering program , an image filtering method, and an ultrasonic diagnostic apparatus are provided.

  That is, the present invention is an ultrasonic image filtering device that calculates, as an input matrix, each pixel value in a predetermined range with a predetermined pixel of image data as a reference position, and outputs an output value corresponding to the input. Sorting means for sorting each element of each column, and the matrix of the sorting result of the sorting means, the number of rows comprising the row including the reference position and its surrounding rows is less than the number of rows of the input matrix, Extraction means for extracting a submatrix whose number is equal to the number of columns of the input matrix; and calculating means for calculating a median value of each element of the submatrix extracted by the extraction means, and using the median value as the output value; It is characterized by comprising.

  Further, the present invention is an ultrasonic image filtering device that calculates an output value corresponding to the input by using each pixel value in a predetermined range with a predetermined pixel of the image data as a reference position as an input matrix, and each of the input matrix The sorting means for sorting each element of the column for each row, and the sorting result matrix of the sorting means includes a column including the reference position, the number of columns is smaller than the number of columns of the input matrix, and the number of rows is the input. Extraction means for extracting a submatrix equal to the number of rows of the matrix, and calculation means for calculating a median value of each element of the submatrix extracted by the extraction means and using the median value as the output value. It is characterized by that.

  Furthermore, the present invention provides an ultrasonic image filtering device that calculates an output value corresponding to the input by using each pixel value in a predetermined range with a predetermined pixel of the image data as a reference position as an input matrix. A sorting unit that sorts each element in one direction for each element in a second direction orthogonal to the first direction, and a matrix of a sorting result of the sorting unit in the first direction including the reference position. The number of elements in the first direction is less than the number of elements in the first direction of the input matrix, and the number of elements in the second direction is the input. Extraction means for extracting a submatrix equal to the number of elements in the second direction of the matrix, calculation of a median value of each element of the submatrix extracted by the extraction means, and calculation using the median value as the output value And means.

The present invention is a program for executing each pixel value in the predetermined range as a reference position where a predetermined pixel of the image data by the ultrasound as an input matrix, the calculation of the output value corresponding to the input into the computer, the A sorting procedure for sorting each element of each row of the input matrix by column, a first procedure for urging to sort each element of each row of the input matrix by column, and the reference position from the matrix of the sorting result A second procedure for encouraging to extract a submatrix comprising a row including and surrounding rows, wherein the number of rows is less than the number of rows of the input matrix and the number of columns is equal to the number of columns of the input matrix; A program for causing a computer to execute a third procedure for calculating a median value of each element of a matrix and urging the median value to be the output value .

The present invention is an ultrasonic image filtering method for calculating an output value corresponding to the input by using each pixel value in a predetermined range with a predetermined pixel of image data as a reference position as an input matrix, and for each row of the input matrix A first step of sorting each element by column, and the matrix of the sorting result is composed of a row including the reference position and its surrounding rows, the number of rows is less than the number of rows of the input matrix, and the number of columns is A second step of extracting a sub-matrix equal to the number of columns of the input matrix, and a third step of calculating a median value of each element of the sub-matrix and using the median value as the output value. It is characterized by.
The present invention generates means for generating image data based on a received signal obtained by transmission / reception of ultrasonic waves, and generates an input matrix having pixel values within a predetermined range including a pixel at a reference position as element values from the image data. And a sorting means for sorting the element values of the input matrix for each column or row, and a sub-matrix that includes the reference position and is smaller than the predetermined range is extracted from the matrix sorted by the sorting means Extracting means for selecting, a selecting means for selecting a median value from a plurality of element values constituting the extracted sub-matrix, and a filtering image generating means for generating a filtered image composed of the median values selected by the selecting means And a display means for displaying the filtered image.
The present invention comprises means for generating an input matrix having pixel values within a predetermined range including a pixel at a reference position from image data as element values, and sorting means for sorting the element values of the input matrix for each column or row. Extracting means for extracting a submatrix that includes the reference position and is smaller than the predetermined range from the matrix sorted by the sorting means, and values of a plurality of elements constituting the submatrix extracted by the extracting means An image filtering apparatus comprising: selecting means for selecting a median from the image data.
The present invention is a procedure for generating an input matrix having pixel values within a predetermined range including pixels at a reference position from image data as element values, a procedure for sorting the element values of the input matrix for each column or row, A procedure for extracting a submatrix that includes the reference position and is smaller than the predetermined range from the matrix sorted by the sorting means, and selects a median value from values of a plurality of elements constituting the extracted submatrix This is a program for causing a computer to execute the procedure to be performed.
The present invention generates an input matrix whose element values are pixel values within a predetermined range including a pixel at a reference position from image data, and sorts the element values of the input matrix for each column or row, Extracting a submatrix including the reference position and having a range smaller than the predetermined range from the matrix sorted by the sorting means, and selecting a median value from values of a plurality of elements constituting the extracted submatrix And an image filtering method.

According to the present invention, an image filtering device , an image filtering program , an image filtering method, and an ultrasonic diagnosis that have the same effects as the median filter, such as smoothing and noise reduction image omission reduction, and are faster than the median filter. An apparatus can be provided.

  Embodiments of the present invention will be described below with reference to the drawings.

(First embodiment)
Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 is a block diagram showing the overall configuration of the ultrasonic diagnostic apparatus according to the first embodiment of the present invention.

  In FIG. 1, this ultrasonic diagnostic apparatus 10 is for performing ultrasonic transmission / reception with respect to a subject (not shown) and ultrasonic transmission / reception with respect to a predetermined scanning direction. A transmission / reception unit 12 that transmits / receives electrical signals to / from the ultrasonic probe 11, an image processing unit 13 that generates and processes B-mode ultrasonic image data based on a reception ultrasonic signal obtained from a predetermined scanning direction, And a display unit 16 for displaying the ultrasonic image processed by the image processing unit 13.

  The image processing unit 13 includes a B-mode image processing unit 14 that generates and processes B-mode image data from a received ultrasonic signal, and a Doppler image processing unit 15 that generates and processes Doppler image data from a received ultrasonic signal. It has. The display unit 16 also converts a scanning method of the ultrasonic image processed by the image processing unit 13 for display, and a color for displaying the ultrasonic image scanned and converted by the scan converter 17. It consists of a monitor 18.

  In the ultrasonic diagnostic apparatus 10 having such a configuration, an ultrasonic beam is irradiated in a predetermined operation direction from the ultrasonic probe 11 to the subject (not shown) via the transmission / reception unit 12. When receiving the reflected beam, a reception signal (reception echo) is output from the ultrasonic probe 11 to the transmission / reception unit 12. Then, based on the received ultrasonic signal received by the transmission / reception unit 12, the B-mode image processing unit 14 in the image processing unit 13 generates and processes B-mode image data, and the Doppler image processing unit 15. The Doppler image data is generated and processed at.

  Then, the generated and processed B-mode image and Doppler image data are converted into a display operation method by the scan converter 17 in the display unit 16. Next, the ultrasonic image scanned and converted by the scan converter 17 is displayed on the color monitor 18.

  Next, the configuration of the Doppler image processing unit 15 shown in FIG. 1 will be described in detail with reference to FIG.

  FIG. 2 is a block diagram illustrating a detailed configuration of the Doppler image processing unit 15. The Doppler image processing unit 15 includes a quadrature detector 21, an A / D converter 22, a Doppler signal storage unit 23, an MTI (Moving Target Indication) filter 24, an autocorrelator 25, and a calculator 26. A non-linear spatial filter (sorting filter) 28 and a linear spatial filter (smooth filter) 29.

  The quadrature detector 21 performs quadrature detection on a received signal obtained by a plurality of times of ultrasonic transmission / reception, and outputs it as an IQ signal having I and Q components that are 90 degrees different in phase. The IQ signal is converted into a digital signal by the A / D converter 22 and sequentially stored in the Doppler signal storage unit 23. In the Doppler signal storage unit 23, IQ signals stored in the order of positions (depths) in the same scanning direction are read out in the order of transmission / reception times of the same positions.

  In the MTI filter 24, components (clutter signal components) caused by the movement of the living tissue are removed from the IQ signal read out by the Doppler signal storage unit 23, and only blood flow information is extracted. The autocorrelator 25 performs autocorrelation processing on the Doppler signal that has passed through the MTI filter 24. Based on the autocorrelation processing result of the autocorrelator 25, the computing unit 27 calculates an average blood flow velocity value, a variance value, a power value, and the like, and generates color Doppler image data.

  The color Doppler image data output from the computing unit 27 is subjected to filtering processing by the sorting filter 28 and the smooth filter 29 and output to the display unit 16. The sorting filter 28 smoothes the boundary lines of blood vessels and blood flow distributions, removes noise, and reduces image omission. The sorting filter 28 is a main part in the present embodiment, and the detailed processing procedure will be described later.

  The sorting filter 28 corresponds to a median filter in the prior art. The smooth filter 29 is also for smoothing the image, but its transfer function is generally linear, and the filtering effect is different from that of the sorting filter 28.

  In this embodiment, the processing is performed in the order of the sorting filter 28 and the smooth filter 29. However, the order of these processing is not limited, and the smooth filter 29 may be first. In this embodiment, it is assumed that the processing by the sorting filter is performed only on the speed value image data among the color Doppler image data. However, dispersion value image data and power value image data may be used.

  Next, the processing procedure of the sorting filter 28 will be described with reference to FIGS.

  FIG. 3 is a diagram showing velocity value image data at a predetermined time as a matrix in the scanning direction and the depth direction.

  A predetermined range having the predetermined position 31 as a reference position is set as an input matrix 32 for filtering the predetermined position 31 in the matrix 30 of the entire velocity value image data. FIG. 4A is a diagram illustrating an example of each pixel value of the input matrix 32. As a first processing procedure (first stage (first step)) of the sorting filter 28, sorting is performed for each column of the input matrix 32 based on the magnitude of the value. In the present embodiment, all sorts are described in ascending order. However, even if all of these are read as “descending order”, the same result is obtained.

  An example of each pixel value after completion of the first processing procedure described above is shown in FIG. 4B as a 7 × 3 matrix 32a. In the figure, each row is represented as 33a, 33b and 33c and surrounded by a frame as shown.

  Next, as a second processing procedure (second stage (second step)) of the sorting filter 28, a partial matrix 34 is extracted from the 7 × 3 matrix 32a sorted by the first processing procedure. The In the present embodiment, as shown in FIG. 4C, a total of three rows, that is, the row 35b at the reference position 31 and the rows 35a and 35c on both sides thereof are extracted. As a result, the submatrix 34 has 3 rows and 3 columns.

  By the way, in this embodiment, the number of matrices of the input matrix 32 is 7 rows and 3 columns, but the number of matrices of the input matrix in the present invention is not necessarily the same. For example, the input matrix may be 3 rows and 3 columns, or 5 rows and 5 columns. However, it is desirable that the number of rows and columns of the input matrix 32 is an odd number. Further, the number of columns of the submatrix 34 needs to be the same as that of the input matrix 32, but the number of rows may be smaller than that of the input matrix 32. Note that the embodiments of 3 rows and 3 columns and 5 rows and 5 columns will be described later as third and fourth embodiments.

  Next, as a third processing procedure of the sorting filter 28, the median value (median value) of the submatrix 34 is used as the output value of the sorting filter 28. However, since the sorting for each column has already been completed at the time shown in FIG. 4 (c), in this processing procedure, the following two steps may be performed.

  As the first stage of the third processing procedure (stage 3-1 (third step)), sorting is performed for each row of the submatrix 34. FIG. 4D is a diagram illustrating an example of each pixel value after the completion of the third step.

  Then, in the next stage (stage 3-2 (fourth step)) of the third processing procedure, the first row of the first matrix is derived from the submatrix 34a sorted in the first stage (third step). The third column, the second row, the second column, and the third row, first column are extracted, and the pixels at other positions are removed. That is, as shown in FIG. 4E, the pixels 37a and 37b are removed while leaving the pixel 36 including the reference position 31. It is logical that pixel values at positions other than the first row, third column, second row, second column, and third row, first column are not median values at the end of the fourth step described above. It is clear.

  Further, if the pixel set 62 composed of three pixels is sorted to obtain the pixel set 36a shown in FIG. 4F, the pixel at the reference position 31 is output.

  The similarity between the filter and the median filter in this embodiment can be calculated as follows.

First, the probability that the result of the filter in this embodiment matches the result of the median filter having the same input matrix size is calculated. The conditions under which the results of the two filters match are as follows.
(I) The position of the pixel indicating the median value of the input matrix 32 (“25” in this embodiment) is in the submatrix 34, that is, 3 of the matrix 32a when the first step is completed. Being between line 5 or 5.
(Ii) In the submatrix 34, the number of pixels smaller than the output value of the filter is equal to the number of larger pixels. This is because the output value of the filter in this embodiment is the median value of the submatrix 34.

For example, the input matrix values in this embodiment satisfy the above (i) and (ii), and both the filter and the median filter in this embodiment give the result “25”. In the probability calculation, pixels smaller than the output value of the filter are group A (number of elements N), and pixels larger than the output value are group B. Under the above conditions, the following expression (1) is established.

However, P ₁ selects each pixel of the column including the output value of the filter (excluding the output value of the filter itself) from groups A and B, and the probability when the number of groups A is x
P ₂ is the probability when each pixel in another column is selected from the remaining pixels, and the number of groups A is y.

At this time, P ₁ and P ₂ are expressed by the following equations (2) and (3).

Since the output value of the filter is equal to the median value, the number N of elements in group A is 10. Therefore, the combination of x and y and P1, P2, and P under the above conditions are as shown in Table 1 below.

  In other words, the probability that the filter result in this embodiment matches the median filter result is 0.8091.

  Next, the probability that the result of the filter in the present embodiment differs from the result of the median filter by only one order of sorting is calculated.

  The probability calculation method uses the above formulas (1) and (2), and N = 11 or 9. Here, if the calculation process is omitted and only the results are shown, P = 0.0804 for N = 11 and N = 9, respectively.

  From the above, the probability that the filter result in the present embodiment differs from the median filter result and the sort order at most by only one before and after is 0.8091 + 2 × 0.0804 = 0.97. It can be concluded that the filter in form gives almost the same result as the median filter.

  Next, in order to show that the present embodiment has a smaller number of calculations than the complete median filter and can be processed at high speed, a calculation method for a complete median filter of 7 rows and 3 columns will be described below with reference to FIG.

The procedure for sorting the original input matrix 32 column by column to obtain the matrix 32a is the same as in this embodiment, but the subsequent processing is different.
That is, the entire matrix 32a as shown in FIG. 5 (a) is then sorted by row to obtain the matrix 32b as shown in FIG. 5 (b). The state of numerical values before and after sorting is shown in FIGS. 5 (a) and 5 (b). Each row to be sorted is represented by 61a to 61g and surrounded by a frame as shown in the figure.

  Each pixel represented by 62 shown in FIG. 5C is extracted from the matrix 32b. Then, the other positions, that is, the pixels represented by 63a and 63b shown in FIG. 5C are not medians and are removed. The position of each row of the pixels 62 is reconstructed to obtain a matrix 62a as shown in FIG.

  Thereafter, if a process similar to that of the submatrix 34 in the present embodiment is performed, a complete median of the input matrix 32 can be obtained at the predetermined position 31.

  The number of calculations of the complete median filter of the above 7 rows and 3 columns is calculated as the number of comparison / replacement of 2 pixels.

Sorting 3 or 7 data by the merge = replacement method requires 3 and 16 comparisons, respectively. Therefore, the number of comparisons required at each stage of processing and the total
Sort by 7 rows and 3 columns ... 48 times
7 rows by 3 columns by row sorting 21 times
Sort by 3 rows by 3 columns ... 9 times
3 pixel sorting ... 3 times
In contrast to this, the number of times of the calculation of the 7 × 3 filter in this embodiment is 81 times.
Sort by 7 rows and 3 columns ... 48 times
Sort by 3 rows by 3 columns ... 9 times
3 pixel sorting ... 3 times
60 times in total
Thus, the number of calculations is reduced.

  Therefore, according to the present embodiment, it is possible to perform high-speed image filtering with the same number of calculations as the median filter while having the same effect as the median filter.

(Second Embodiment)
In the first embodiment described above, first, each element of each row of the input matrix 32 is sorted by column, and then a partial matrix composed of the row including the reference position and surrounding rows is extracted from the matrix of the sorting result. However, it is not limited to this. For example, in the second embodiment, each element of each column of the input matrix 32 is sorted for each row.

  Hereinafter, the processing procedure of the sorting filter 28 in the second embodiment of the present invention will be described with reference to FIG.

  FIG. 3 is a diagram showing velocity value image data at a predetermined time as a matrix in the scanning direction and the depth direction.

  FIG. 6A is a diagram illustrating an example of each pixel value of the input matrix 42. As a first step of the sorting filter 28, sorting is performed for each row of the input matrix 32. An example of each pixel value after completion of the first step described above is shown in FIG. 6B as a 7 × 3 matrix 42a. In the figure, each row is represented by 43a to 43g and surrounded by a frame as shown.

  Next, as a second step of the sorting filter 28, a submatrix 44 is extracted from the 7 × 3 matrix 42a sorted by the first step. In the present embodiment, as shown in FIG. 6C, one column 45 of the reference position 41 is extracted. As a result, the submatrix 44 has 7 rows and 1 column.

  Next, as a third step of the sorting filter 28, the columns of the submatrix 44 are sorted. FIG. 6D is a diagram illustrating an example of each pixel value after the completion of the third step.

  Then, in the fourth step, as shown in FIG. 6E, the pixel 46 including the reference position 47 is left and other pixels are removed from the partial matrix 44a sorted in the third step.

The number of calculations by the above processing method is as follows.
7 rows by 3 columns by row sorting 21 times
Sorting 7 rows 1 column 16 times
38 times
Thus, the number of calculations is reduced compared to a complete median filter of 7 rows and 3 columns. However, since the filter characteristics and the similarity with the median filter are different from those of the first embodiment described above, it is necessary to confirm the image quality separately from the first embodiment.

  Thus, even if the sorting order is changed, the same effect as that of the first embodiment described above can be obtained.

  In the second embodiment, as in the first embodiment described above, all sorts are described in ascending order. However, even if these are all set in “descending order”, the same result is obtained. Furthermore, it is needless to say that the number of input matrices in this embodiment is not necessarily the same.

(Third embodiment)
In the first and second embodiments described above, a 7 × 3 sorting filter has been described as an example. In the third embodiment, a 3 × 3 sorting filter having a size often mentioned in the median filter embodiment will be described.

  Hereinafter, the processing procedure of the 3 × 3 (3 × 3) sorting filter will be described with reference to FIG.

  FIG. 7A is a diagram illustrating an example of each pixel value in the predetermined range 51. Sorting is performed for each predetermined range, that is, for each column of the input matrix 51 by the same method as in the above-described embodiment. FIG. 7B is a diagram illustrating an example of each pixel value after the end of the first processing procedure. In addition, each column is expressed as a frame 53a, 53b and 53c.

  Next, as shown in FIG. 7C, the second row, that is, 54 is extracted from the 3 × 3 matrix 52a sorted for each column. Further, as shown in FIG. 7D, the pixels 54a are sorted, and their median value (median value) is set as the output value 55 of the sorting filter.

  In the conventional example for the median filter, the processing procedure is the order of 3 column sorting, 3 row sorting, and 3 pixel sorting. In the third embodiment, the 3 column sorting and 1 row sorting are performed. In other words, it is apparent that the number of calculations is smaller in the present embodiment than in the median filter at the same size.

  Thus, according to the above-described embodiment, it is possible to provide a high-speed image filtering method that requires fewer calculations than the median filter.

(Fourth embodiment)
Next, a fourth embodiment of the present invention will be described.

  In the above-described third embodiment, a 3 × 3 square matrix has been described. Even if the input matrix is a square matrix, a 5 × 5 filter is a more complicated process. Hereinafter, the processing procedure of the sorting filter 28 according to the fourth embodiment of the present invention will be described with reference to FIG.

  FIG. 8A is a diagram illustrating an example of each pixel value of the input matrix 72 around the predetermined position 71. As the first processing procedure (first stage (first step)) of the sorting filter 28, sorting is performed for each column of the input matrix 72 based on the magnitude of the value. In the present embodiment, all sorts are described in ascending order. However, even if all of these are read as “descending order”, the same result is obtained.

  FIG. 8B shows an example of each pixel value after completion of the first processing procedure described above as a 5-by-5 matrix 72a. In the figure, each row is represented by 73a to 73e and surrounded by a frame as shown.

  Next, as a second processing procedure (second stage (second step)) of the sorting filter 28, a sub-matrix 74 is extracted from the 5-by-5 matrix 72a sorted by the first processing procedure. The In the present embodiment, as shown in FIG. 8C, a total of three rows, that is, the row 75b at the reference position 71 and the rows 75a and 75c existing on both sides of the row 75b are extracted. As a result, the submatrix 74 has 3 rows and 5 columns.

  Next, as a third processing procedure of the sorting filter 28, the median value (median value) of the submatrix 74 is used as the output value of the sorting filter 28. However, since the sorting for each column has already been completed at the time shown in FIG. 8 (c), in this processing procedure, the following two steps may be performed.

  As the first stage of the third processing procedure (stage 3-1 (third step)), sorting is performed for each row of the submatrix 74. FIG. 8D is a diagram illustrating an example of each pixel value after the completion of the third step.

  Then, in the next stage (third stage (second step) (fourth step)) of the third processing procedure, the first row of the first row from the submatrix 74a sorted in the first stage (third step). The fifth column, the first row, the fourth column, the second row, the third column, the third row, the second column, and the third row, the first column are extracted, and pixels at other positions are removed. That is, as shown in FIG. 8E, a pixel set 76 including the reference position 71 is left, and pixel sets 77a and 77b that are clearly not median values are removed.

  Further, if the pixel set 76 composed of these five pixels is sorted to obtain a pixel set 76a as shown in FIG. 8F, the pixel at the reference position 71 is output.

  Next, as in the description of the first embodiment, in order to compare the number of calculations between this embodiment and a complete median filter having the same input matrix size, first, a complete 5 × 5 complete median filter is used. The calculation method will be described below with reference to FIG. This method is based on the method described on page 8 and after of “Fast Algorithms for Median Filtering”, Intel Corporation, 2001.

  The procedure of sorting the original input matrix 72 for each column to obtain the matrix 72a is the same as that in the present embodiment, but the subsequent processing is different.

  That is, the entire matrix 72a as shown in FIG. 9 (a) is then sorted by row to obtain the matrix 72b as shown in FIG. 9 (b). The state of the numerical values before and after sorting is as shown in FIGS. 9A and 9B, and each row to be sorted is represented by 81a to 81e and surrounded by a frame as shown.

  Next, each pixel of the matrix 72b is classified into the following four as shown in FIG. However, the order in parentheses is (column, row).

82a (1, 4), (2, 3), (3, 2), (4, 1)
82b (1,5), (2,4), (3,3), (4,2), (5,1)
82c (2,5), (3,4), (4,3), (5,2)
82d... Pixel other than the above 82d is not a median and is removed, and the minimum value of 82a (in this case "18"), the median value of 82b (in this case "19"), the maximum value of 82c (in this case " The median value of the three pixels obtained by taking 21 ″) is the complete median of the original input matrix 72.

The number of calculations of the complete median filter of 5 rows and 5 columns is as follows. In order to sort 3, 5 data by the merge = replacement method, 3 and 9 comparisons are required, respectively.
Sort by 5 rows and 5 columns: 45 times
Sorting every 5 rows by 5 columns ... 45 times
Maximum selection of 4 pixels ... 3 times
4 pixel minimum value selection ... 3 times
5 pixel sorting ... 9 times
3 pixel sorting ... 3 times
In contrast, the number of calculations of the filter according to the present embodiment is as follows, and the number of calculations is less than that of a complete median filter.
Sort by 5 rows and 5 columns: 45 times
Sort by 3 rows by 5 columns ... 27 times
5 pixel sorting ... 9 times
81 times in total As described above, according to the embodiment described above, it is possible to provide a high-speed image filtering method that requires fewer calculations than the median filter.

  As mentioned above, although embodiment of this invention was described, in the range which does not deviate from the summary of this invention other than embodiment mentioned above, this invention can be variously modified.

  Further, the above-described embodiments include inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some constituent requirements are deleted from all the constituent requirements shown in the embodiment, the problem described in the column of the problem to be solved by the invention can be solved, and the effect described in the column of the effect of the invention Can be extracted as an invention.

1 is a block diagram showing an overall configuration of an ultrasonic diagnostic apparatus according to a first embodiment of the present invention. FIG. 2 is a block diagram illustrating a detailed configuration of a Doppler image processing unit 15 in FIG. 1. FIG. 6 is a diagram illustrating a processing procedure of the sorting filter 28, and is a diagram showing velocity value image data at a predetermined time as a matrix in the scanning direction and the depth direction. FIG. 10 is a diagram illustrating a processing procedure of the sorting filter and is a diagram illustrating an example of each pixel value after the processing procedure is completed. FIG. 11 is a diagram illustrating another processing procedure of the sorting filter 28 and is a diagram illustrating an example of each pixel value after the processing procedure is completed according to a 7-by-3 complete median filter calculation method. It is a figure for demonstrating the process sequence of the sorting filter 28 in the 2nd Embodiment of this invention. It is a figure for demonstrating the process sequence of the sorting filter 28 in the 3rd Embodiment of this invention. It is a figure for demonstrating the process sequence of the sorting filter 28 in the 4th Embodiment of this invention. FIG. 10 is a diagram illustrating a processing procedure of the sorting filter in the fourth embodiment of the present invention, and is a diagram illustrating an example of each pixel value after completion of the processing procedure according to a calculation method of a complete median filter of 5 rows and 5 columns. It is.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Ultrasound diagnostic apparatus, 11 ... Ultrasonic probe, 12 ... Transmission / reception part, 13 ... Image processing part, 14 ... B mode image processing part, 15 ... Doppler image processing part, 16 ... Display part, 17 ... Scan converter, 18 DESCRIPTION OF SYMBOLS ... Color monitor, 21 ... Quadrature phase detector, 22 ... A / D converter, 23 ... Doppler signal memory | storage part, 24 ... MTI (Moving Target Indication) filter, 25 ... Autocorrelator, 26 ... Calculator, 28 ... Nonlinear Spatial filter (sorting filter), 29 ... linear spatial filter (smooth filter), 30 ... matrix, 31 ... predetermined position (reference position), 32 ... input matrix.

Claims (17)

An ultrasonic image filtering device that calculates an output value corresponding to the input, with each pixel value in a predetermined range having a predetermined pixel of image data as a reference position as an input matrix,
Sorting means for sorting each element of each row of the input matrix by column;
A sub-matrix consisting of a row including the reference position and surrounding rows from the matrix of the sorting result of the sorting means, wherein the number of rows is less than the number of rows of the input matrix and the number of columns is equal to the number of columns of the input matrix. Extracting means for extracting;
Calculating a median value of each element of the submatrix extracted by the extraction unit, and calculating the median as the output value;
An image filtering apparatus comprising:   The calculation means sorts each element of each column of the submatrix for each row, and it is logically known in advance from the sorted submatrix that it is not the median value of each element of the submatrix. The image filtering apparatus according to claim 1, wherein an element at a predetermined position is excluded. An ultrasonic image filtering device that calculates an output value corresponding to the input, with each pixel value in a predetermined range having a predetermined pixel of image data as a reference position as an input matrix,
Sorting means for sorting each element of each column of the input matrix for each row;
Extraction means for extracting a sub-matrix composed of columns including the reference position, the number of columns being less than the number of columns of the input matrix, and the number of rows being equal to the number of rows of the input matrix, from the matrix of the sorting result of the sorting means ,
Calculating a median value of each element of the submatrix extracted by the extraction unit, and calculating the median as the output value;
An image filtering apparatus comprising:   The extraction means sorts each element of each row of the submatrix for each column, and from the sorted submatrix, it is logically known beforehand that it is not the median value of each element of the submatrix. The image filtering apparatus according to claim 3, wherein a position element is excluded. An ultrasonic image filtering device that calculates an output value corresponding to the input, with each pixel value in a predetermined range having a predetermined pixel of image data as a reference position as an input matrix,
Sorting means for sorting each element in the first direction of the input matrix for each element in a second direction orthogonal to the first direction;
From the matrix of the sorting results of the sorting means, the number of elements in the first direction, which is composed of the elements in the first direction including the reference position and the elements in the first direction around it, is the first of the input matrix. Extracting means for extracting a sub-matrix that is less than the number of elements in the second direction and whose number of elements in the second direction is equal to the number of elements in the second direction of the input matrix;
Calculating a median value of each element of the submatrix extracted by the extraction unit, and calculating the median as the output value;
An image filtering apparatus comprising:   The calculation means sorts the elements in the second direction of the submatrix for each element in the first direction, and determines in advance from the sorted submatrix that it is not the median value of each element of the submatrix. The image filtering apparatus according to claim 5, wherein an element at a predetermined position that is logically known is excluded.   The image filtering apparatus according to claim 5 or 6, wherein the first direction is a row direction, and the second direction is a column direction.   The image filtering apparatus according to claim 5, wherein the first direction is a column direction, and the second direction is a row direction. A program for executing each pixel value in the predetermined range as a reference position where a predetermined pixel of the image data by the ultrasound as an input matrix, the calculation of the output value corresponding to the input into the computer,
A sorting procedure for sorting each element of each row of the input matrix by column;
A first procedure that prompts the elements of each row of the input matrix to be sorted by column;
A sub-matrix consisting of a row including the reference position and surrounding rows is extracted from the sorting result matrix, and the number of rows is smaller than the number of rows of the input matrix and the number of columns is equal to the number of columns of the input matrix. A second procedure that prompts
Calculating a median value of each element of the submatrix and prompting the median value to be the output value;
A program that causes a computer to execute. As the third procedure , a 3-1 procedure for urging to sort each element of each column of the submatrix by row;
3-2 urging to exclude an element at a predetermined position that is logically known in advance not to be the median value of each element of the submatrix from the submatrix obtained by the sorting according to the step 3-1 And the steps
Program according to claim 9 for causing the computer to perform the. An ultrasonic image filtering method for calculating an output value corresponding to the input, with each pixel value in a predetermined range having a predetermined pixel of image data as a reference position as an input matrix,
A first step of sorting each element of each row of the input matrix by column;
A sub-matrix that includes a row including the reference position and surrounding rows is extracted from the matrix of the sorting result, the number of rows being less than the number of rows of the input matrix and the number of columns being equal to the number of columns of the input matrix. Two stages,
Calculating a median value of each element of the submatrix, and using the median value as the output value;
An image filtering method comprising: In the third step, the third step of sorting each element of each column of the submatrix by row,
A step 3-2 for excluding an element at a predetermined position that is logically known in advance not to be a median value of each element of the submatrix from the submatrix sorted in the step 3-1. ,
The image filtering method according to claim 11, further comprising: Means for generating image data based on a reception signal obtained by transmission and reception of ultrasonic waves;
  Means for generating an input matrix having a pixel value within a predetermined range including a pixel at a reference position as an element value from the image data;
Sorting means for sorting the element values of the input matrix by column or row;
Extraction means for extracting a sub-matrix that includes the reference position and is smaller than the predetermined range from the matrix sorted by the sorting means;
Selecting means for selecting a median value from a plurality of element values constituting the extracted submatrix;
Filtering image generation means for generating a filtering image composed of the median value selected by the selection means;
Display means for displaying the filtered image;
An ultrasonic diagnostic apparatus comprising: The ultrasonic diagnostic apparatus according to claim 13, wherein the partial matrix includes an element of the reference position and an element included in a row around the reference position in the matrix sorted by the sorting unit. Means for generating an input matrix whose element value is a pixel value within a predetermined range including a pixel at a reference position from image data;
Sorting means for sorting the element values of the input matrix by column or row;
Extraction means for extracting a sub-matrix that includes the reference position and is smaller than the predetermined range from the matrix sorted by the sorting means;
Selecting means for selecting a median value from values of a plurality of elements constituting the submatrix extracted by the extracting means;
An image filtering apparatus comprising: A procedure for generating an input matrix having element values of pixel values within a predetermined range including a pixel at a reference position from image data;
Sorting the element values of the input matrix by column or row;
A procedure for extracting a sub-matrix that includes the reference position and is smaller than the predetermined range from the matrix sorted by the sorting unit;
Selecting a median value from the values of a plurality of elements constituting the extracted submatrix;
A program that causes a computer to execute. Generating an input matrix having pixel values within a predetermined range including the pixel at the reference position from the image data as element values;
Sorting the element values of the input matrix by column or row;
Extracting a sub-matrix that includes the reference position and is smaller than the predetermined range from the matrix sorted by the sorting means;
Selecting a median from values of a plurality of elements constituting the extracted submatrix;
An image filtering method comprising:

Images