JPS6063589A - Data processor - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Technical Field] The present invention relates to a data processing device that creates interpolated data for enlarging and displaying signals based on one-dimensional data, enlarging images based on two-dimensional data, and rotating displays. .

[Prior Art] Conventionally, this type of data processing device simply interpolates data. For example, consider a case where a signal waveform as shown in FIG. 1A is enlarged twice in the horizontal axis direction. First, the signal waveform is sampled and quantized and stored in the first memory as shown in Figure 1 (B), and then the data string in Figure 1 (B) is expanded twice on the horizontal axis and stored in the second memory. stored in memory. FIG. 1(C) shows this state. Therefore, the contents of the second memory are in a state where data is omitted every other memory as a result of the data being expanded and stored. Therefore, it is necessary to interpolate between this period, but according to the conventional method, an interpolation method is used for data between this period, for example, in which the average value of data on both sides is taken. Figure 1 (
D) is the result of signal enlargement display by the interpolation described above, and the white circles are data points created by interpolation. The dashed line a shows the ideal expanded state of the original data; compared to this, when the number of quantization points is small, the sharply changing parts of the signal become smoother due to interpolation. was. Furthermore, in order to alleviate the above-mentioned drawbacks, an interpolation method using high-order calculations is required, which complicates the processing.

[Objective] It is an object of the present invention to eliminate the drawbacks of the above-mentioned prior art and to propose a data processing device that can perform data interpolation processing without losing the characteristics of a sharply changing portion of a signal.

[Examples] Examples according to the present invention will be described in detail below with reference to the drawings. FIG. 2 is a block diagram showing the functional configuration of data interpolation processing of the data processing device of the embodiment. In the figure, 1 is a microprocessor, and its interior is shown as a combination of functions executed by software. 2 to 6 are shift registers that temporarily hold a part of the sampling data string and a part of the interpolation data, and are composed of memories. ),
(i), (i+1) (() refers to the content, the same applies hereinafter) are stored, and 5 stores sampling data (i-1) and (i
) is stored, and the interpolated data (i) of register 3 is stored at the next timing.
' (currently blank b) is determined. For the relationship between these sampling data and interpolation data, please refer to the waveform diagram in Fig. 3. Arrow P indicates the direction of progress of the processing procedure. It is determined whether the size is larger than a predetermined value α.

The discrimination output drives the selection means 10. 8 is an interpolation means which interpolates internally interpolated data (i)'=((i)+(1+1))/ from sampling data (i) and (i+1)
Get 2. 9 is an extrapolation means and sampling data (i
-1) and (i), and from data (i-1) to (i
), put the interpolated data (i)' on the extension of the straight line. In the embodiment, this is called extrapolation, and the selection means 10 selects and outputs the output of the extrapolation means 9 when the discrimination output of the discrimination means 7 is 1 (large), and when the discrimination output is O (no). At this time, the output of the interpolation means 8 is selected and outputted. The selected output contents are stored in the register 3.

FIG. 4 shows a processing procedure executed by software, and an interrupt is input to this flow every time sampling data is sampled. Of course, all the sampling data may be stored in the RAM in advance and interpolated data may be obtained by scanning this data using software.In step 40, the latest sampling data (I+1) is shifted into register 2. Ru. In the previous state, sampling data (f) was stored in register 2.

Therefore, when the latest data (i+B) is shifted in, data (i) is shifted twice and transferred to register 4, and blank code b is inserted into register 3. In other words,
The contents of register 3 are now determined by interpolation processing. In step 41, data (i) and data (i
+1) and its absolute value is compared with a preset fixed value α to determine the magnitude relationship. If the determination exceeds the fixed value α, it is determined that the area containing the interpolated data to be created is an area where the signal changes sharply, and the process proceeds to step 42, where an extrapolation procedure is performed. Also,
If the determination in step 41 is NO, it is determined that the area containing the interpolation data to be created is an area where the signal changes slowly, and the process proceeds to step 44, where interpolation procedures are performed. Referring to Figure 3, the extrapolation procedure creates interpolated data (X)' of data (i) and data (i+1) as values on a straight line passing through data (i-1) and data (i). do. On the other hand, to explain the interpolation procedure with reference to the previous data, data (i-1) and data (i) are expressed as values on a straight line passing through data (i-1) and data (i). Interpolated data (i-1)' is created. In step 43, interpolation data is created by executing a process of 0 or more to set the interpolation data obtained by interpolation or extrapolation in register 3 at all points to be interpolated, and the obtained result is The signal waveform of is shown in FIG. According to the data scanning direction P, the black circles are sampling data.

The white circles indicate the interpolated data, and the double circles indicate the extrapolated data. The waveform connecting these values with a straight line is the signal after interpolation, and the waveform shown by the broken line is the original. Corresponds to a signal. Here, unlike the conventional example shown in FIG. 1(D), extrapolation (i)' is performed, so that the steep components of the original signal are not lost. Furthermore, if such processing is performed on a two-dimensional data array, data interpolation processing can be performed on a two-dimensional image signal.

Next, a case will be described in which a filter is applied to the enlarged data that has been subjected to data interpolation. FIG. 6(A) shows a functional block diagram showing this relationship. The filter means 2 of the embodiment can be realized by the microprocessor 1, and filters the enlarged data interpolated by the interpolation means 11. The relationship is shown. The filter means 12 is a filter that emphasizes steep edges of signals, and is often used in processing image data. Here, as a simple example, it is assumed that the filter effect is applied only when the difference between two points is larger than a certain threshold value β. FIG. 5 shows the filter processing procedure of the embodiment. In step 50, input data is shifted in. That is, 1 input data is shifted to register R(j+1) and at the same time, the contents of register R(j+1) are shifted to register 'R(j). Input data is assumed to be a positive value for ease of processing. Such an assumption can be easily realized by applying a uniform bias to the input data. Step 51
Then, from the contents of register R(j), register R(j+1)
The content of is subtracted, and the magnitude relationship with the threshold value β is determined.

If the determination is R(j)-R(J+1)>β, it indicates that a steep edge portion of the falling signal has been detected, and the process proceeds to step 52 to perform edge enhancement processing. So step 5
Process 2 is a process that increases the contents of register R(j) by 10% and decreases the contents of register R(j+1) by 10%. In step 53, it is determined whether the input data is complete. If the data is not completed, return to step 5o and shift in one second data. Again in step 51, the content of register R(j+1) is subtracted from the content of register R(j) to determine whether it is larger than the threshold value β.The determination in step 51 is R(j)−R(j+1)<−β If , it indicates the detection of a steep edge part of the rising signal, and the process proceeds to step 54 to perform edge part emphasizing processing.
This process increases the contents of register R(j+1) by 10%. In step 53, it is similarly determined whether or not the data has ended. Also, the determination in step 51 is IR (
When D-R(j+1)I≦β, there is no steepness in the signal, and edge enhancement processing is not performed.Based on this determination, all data including interpolated data is filtered. At step 53, it is determined that the data has ended and the processing ends.

The signal waveform in FIG. 7(A) is the signal waveform in FIG. 1(D), that is, the data expanded only by interpolation, which is a conventional method, is filtered, and is the part where the signal changes most rapidly. is rounded by interpolation, and this change is smaller than the threshold value β, indicating that no filter effect appears. If the threshold value β is set to a small value, a filtering effect will appear as shown by the broken line, but there is a risk that unnecessary parts will be affected by the filter, making it difficult to select the threshold value. Figure 7(B) shows the signal waveform of Figure 3 expanded by data interpolation using the interpolation method of the present invention, that is, a combination of interpolation and extrapolation, and is filtered to reduce the steepness of the signal. It has been preserved and the effect of the filter is clearly visible.

4. FIGS. 8A to 8D show signal waveforms obtained when the original data signal is first filtered and then enlarged and data interpolated.

A functional block diagram showing the relationship between these processing steps is shown in Figure 6 (
B). FIG. 8(A) is the original signal, and FIG. 8(B) is the signal after filter processing. Figure 8(C) shows the case where data expansion processing is performed on the filtered signal only by interpolation, which is the conventional method, and Figure 8(D)
shows a signal when enlarged processing is performed by data interpolation using an interpolation method according to the present invention, that is, a combination of an interpolation method and an extrapolation method. Only the interpolation points are shown here. In other words, white circles are interpolation points, and double circles are extrapolation points. As can be clearly seen by comparing FIGS. 8(C) and (D), the present invention exhibits good interpolation characteristics even for data after filtering, that is, the steepness of signal 5 is lost. It won't happen.

In the description of the above embodiment, the fixed value α prepared in advance for comparison in step 41 of FIG. 4 should be determined by the sampling frequency when quantizing data and the characteristics of the signal waveform to be handled. When used in combination with a filter, it is important to consider the fixed value α and the characteristics of the filter. For example, when applying a filter such as the edge enhancement described above, the characteristics of this filter are A delay-type low-pass filter is required, and the filter characteristics should be given so that its resonance point corresponds to the edge portion of the signal that is desired to be emphasized. - Old book Since the purpose of data interpolation according to the present invention is to create interpolated data without reducing the sharpness of the edge portion of the signal, the fixed value 6 α prepared in advance for comparison in step 41 is used to It is necessary to set a value that can be detected. Therefore, the characteristics of the low-pass filter may be set so that the frequency of its resonance point operates at the same edge portion as the steepness of the signal detected at a fixed value α.

In addition, a commonly used convolution method may be used for filter processing to detect or emphasize steep edges of a signal. A prepared edge detection table, for example, a data array consisting of -1, 2, -1, is convolved. Specifically, it is a process of taking and adding the products of corresponding data, and a value other than O is obtained in a part where the signal changes sharply, and a value of O is obtained in other parts. Next, interpolation is performed between the points where the adjacent results are O and 0, and between the points where the results are 0 and points that are not 0, and the points between the points that are not 0 and the points that are not O are excluded. This is an interpolation process. Furthermore, when applying a filter to emphasize the edge portion of the signal, the data array may be convolved with an edge emphasis table, for example -1, 3, -1.

In any of the above cases, the spatial distance for processing is set to 1.

In this example, the intermediate value of two points was used for data interpolation, and the linear extension of two points was used for extrapolation. However, other methods such as quadratic interpolation and extrapolation It goes without saying that more accurate data interpolation can be achieved by using the interpolation method, and in the first-order extrapolation method used in this example, the data flow is from left to right as shown in Figure 3. Although we extrapolated a straight line from the right to the left, the effect is the same if we extrapolate a straight line from right to left.

8 [Effects] As explained above, according to the present invention, when performing data interpolation, interpolation and extrapolation are switched and used according to the characteristics of the signal, so that edge portions of the signal that change sharply can be Data can be interpolated without losing its sharpness.

Furthermore, when used in combination with a filter, matching between filter processing and interpolation processing can be obtained by matching the emphasis characteristics of the filter with the threshold value for switching interpolation or extrapolation, and the edge portions of the signal can be matched as before. This prevents the signal from becoming dull and improves the ability to distinguish between steep parts and non-steep parts of the signal.

[Brief explanation of drawings]

FIGS. 1A to 1D are waveform diagrams of signals obtained by a conventional data interpolation method. FIG. 2 is a block diagram showing the functional configuration of a data processing device according to an embodiment of the present invention. 3 is a waveform diagram of a signal interpolated using the configuration shown in FIG. 2; FIG. 4 is a flowchart showing the processing procedure executed by software; FIG. 5 is a flowchart showing the filter processing procedure of the embodiment; Figures (A) and (B) are block diagrams showing the connection relationship between the filter means and interpolation means, and Figures 7 (A) and (B) are waveforms explaining the effect of applying filter processing to data after interpolation processing. Figure, Figure 8 (A) ~
(D) is a waveform diagram illustrating the effect of performing interpolation processing on the filtered original data. Here, 1...microprocessor, 2-6...shift register, 7...comparison means, 8...interpolation means,
9... extrapolation means, lO... selection means, 11... interpolation means, 12... filter means. 0 Figure 3 (i)' Figure 4/Ant; Retained in 1'? 1) ln cane ES 244 Outer god glaze brittle Iη insert ゛Shikihama 4 1 deity 3 Nen offering β Hete one ttsu L- Fig. 7 Fig. 8

Claims (1)

[Scope of Claims] (1) Regarding the arrangement of quantized data, interpolation means interpolates interpolated data of adjacent data, extrapolation means extrapolates the interpolated data, and a discrimination means for discriminating whether the data difference is larger than a set value; and a selection means for selectively outputting the output of the interpolation means or the extrapolation means based on the output of the discrimination means; A data processing device characterized by performing data interpolation. (2) The selection means selectively outputs the output of the extrapolation means when the output of the discrimination means is large, and selectively outputs the output of the interpolation means when the output of the discrimination means is negative. A data processing device according to claim 1. [3] The extrapolation means converts interpolated data (i)' between data (I) and (i+1) into data (i-1) for the data array arranged in the order of (i-1), (1), and (i+1). 2. The data processing device according to claim 1, wherein the data processing device calculates the value on the straight line connecting the points (i). (4) A data processing device comprising a filter means for emphasizing a sharply changing part of quantized data and a data interpolation means for interpolating data between adjacent data, the data interpolation means interpolating the adjacent data. an interpolating means for interpolating data; an extrapolating means for extrapolating the interpolated data; a determining means for determining whether the difference between the adjacent data is greater than a set value; and an output of the determining means. A data processing device comprising a selection means for selectively outputting the output of the interpolation means or the extrapolation means. (5) The filter means has a threshold value for detecting a sharply changing part of the quantized data, and the magnitude of the threshold value is characterized in that it almost matches the magnitude of the setting value of the data interpolation means. A data processing device according to claim 4.