JP3545619B2 - Image arbitrary magnification processing apparatus and method - Google Patents

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image processing apparatus and an image processing method for performing an arbitrary magnification of an image, and more particularly to an apparatus and a method for performing processing such as arbitrary magnification conversion of an image after performing a pan scan (cutting out an image) using MPEG2 or the like. It is about the method.
[0002]
[Prior art]
Recently, in image processing apparatuses, a request has been made to freely enlarge and reduce image information using an arbitrary designated magnification image. On the other hand, as one of the prior arts according to the present invention, there is an "image enlargement / reduction processing apparatus" disclosed in JP-A-10-63826. In this example, processing for enlarging / reducing a video signal is shown. It is.
[0003]
In this conventional example, 1 / N sub-sampling is performed after up-sampling M times when performing any magnification processing.
[0004]
However, a filtering process is required at the time of upsampling by M times and at the time of 1 / N sub-sampling, which tends to increase the circuit scale.
[0005]
[Problems to be solved by the invention]
Therefore, as in the conventional example, when the arbitrary magnification processing of M / N is performed, since the 1 / N sub-sampling processing is performed after the upsampling processing of M times, there is a problem that the circuit scale is increased. Further, in the case of upsampling by M times, since M filter characteristics are required, there is also a problem that the circuit scale is increased.
[0006]
SUMMARY OF THE INVENTION It is an object of the present invention to provide an image processing apparatus that realizes arbitrary magnification processing of an image with a very small filter circuit by performing upsampling processing only on original pixels necessary for arbitrary magnification processing.
[0007]
Further, in the present invention, an arbitrary magnification process can be realized with a simpler configuration by performing a method of approximating an M-fold up-sample to a power-of-two magnification.
[0008]
[Means for Solving the Problems]
In order to solve the above-mentioned problem, the present invention is an image arbitrary magnification processing apparatus that enlarges an original image at an arbitrary magnification m / n, and sets the numerator of the original image at the arbitrary magnification m / n to “16”. In order to express the arbitrary magnification m / n of the original image as “16 / pixel interval specified value”, a first calculating unit for calculating a reference pixel interval base value, and a correction value of the pixel interval specified value. A second calculator for calculating a certain base correction value and a base correction cycle indicating a cycle to be corrected; and a selector for selectively outputting the base correction value calculated by the second calculator based on the base correction cycle. And an adder that receives the pixel interval base value calculated by the first calculator and the base correction value output in the base correction cycle, which is the output of the selector, and adds and outputs these. Receives the output of the adder, A counter output P is calculated by counting the pixel interval base value or the sum of the pixel interval base value and the base correction value with the image cutout start value of the image as an initial value, and indicates a value of 1/16 of the counter output P. A counter for outputting an upper bit and a lower bit indicating a remainder of 1/16 of the counter output P; and an upper bit indicating a value of 1/16 of the counter output P from the counter, and a storage area corresponding thereto is received. And an image memory for reading and outputting the original image from the memory, and a lower bit indicating a remainder of 1/16 of the counter output P from the counter, and filtering the original image read from the image memory accordingly. A coefficient determining unit for determining a coefficient for applying the image data, and the original image read from the image memory is determined by the coefficient determined by the coefficient determining unit. An image any magnification processing apparatus characterized by comprising a multiplier for outputting subjected to Irutaringu process.
[0009]
The present invention relates to an image arbitrary magnification processing apparatus that enlarges given image information at an arbitrary magnification m / n, and fixes the upsampling magnification to “16” which is the easiest to process with a current computer device. Things. This mode is shown in FIG. 3. By fixing the magnification to “16”, it is possible to prepare the filter characteristics of up-sampling that is required for m times in m-times up-sampling only in the case of 16 times. As a result, the circuit configuration can be greatly simplified.
[0022]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[0023]
FIG. 1 is a block diagram showing an example of the first embodiment of the system according to the present invention, and FIG. 2 is an explanatory diagram for explaining the operation of the first embodiment of the system according to the present invention.
[0024]
According to the present invention, when obtaining a converted image obtained by enlarging / reducing each pixel of the original image by m / n times, each pixel constituting the converted image is defined by the nth pixel and the (n + 1) th pixel of the original image. Regarding where it is located, (1) the quotient of P / m → which pixel of the original image, (2) the surplus of P / m → which side is located relative to the two pixels of the original image This is an apparatus for performing an image arbitrary magnification process by obtaining a value or a filter value based on this value. Therefore, the upsampling process is performed only on the necessary pixels of the original image, so that the arbitrary magnification process can be realized with a very small circuit. Hereinafter, a procedure for obtaining two values of the present invention and performing a filtering process based on these values will be described with reference to the drawings. Note that the image arbitrary magnification processing apparatus according to the present invention can be realized based on the same concept as the horizontal processing and the vertical processing of an image. However, in the following embodiment, the horizontal processing will be described.
[0025]
In FIG. 1, a calculating means 11 to which an arbitrary magnification m / n is given is connected to an adder 12, and the combined output is connected to a selector 13, which receives an image cut-out start value and outputs the D- The flip-flop 14 is provided. The output of the D-flip-flop 14 is supplied to a circuit 19 for multiplying the output by 1 / m and a circuit 10 for obtaining a surplus of 1 / m. These outputs are sent to an image memory 15 and a coefficient processor 17 respectively. The output of the image memory 15 can be provided to the filter tap 16. The outputs of the filter tap 16 and the coefficient processor 17 are respectively applied to a multiplier 18 to supply an image output processed at an arbitrary magnification to be obtained.
[0026]
In such a system, first, when the magnification of an image is given from the outside of the apparatus, for example, in the form of m / n, the calculation means 11 of FIG. 1 calculates a pixel interval base value to create an arbitrary magnification image. In this embodiment, the pixel interval base value is n.
[0027]
The pixel interval base value is supplied to the adder 12, and the output of the adder 12 is supplied to the selector 13. The selector 13 normally connects the output of the adder 12 to the D-flip-flop 14. For example, when an initial pulse generated at the start of one horizontal line is applied, a cutout start value of an image given by a pan scan or the like is given. To the D-flip-flop 14.
[0028]
The output of the D-flip-flop 14 is supplied to the filter operation part and also to the adder 12. That is, the adder 12 and the D-flip-flop 14 constitute a counter that counts up with the pixel interval base value with the image cutout start value as an initial value.
[0029]
The count-up output of the D-flip-flop 14 indicates a position indicated by a black triangle in the operation diagram of FIG. Here, the value obtained by dividing 1 / m of the output P of the D-flip-flop 14 indicated by the black triangle mark in the figure, that is, the quotient of P / m, is determined based on which original pixel (FIG. 2 indicates that the two original pixels are normally synthesized at a fixed ratio. The remainder of 1 / m of the output P of the D-flip-flop 14, that is, the remainder of P / m, indicates the ratio of a plurality of corresponding original pixels to be synthesized, and is m-times up-sampled between the original pixels. Shows the value of the filter coefficient for finding the interpolation pixel point.
[0030]
For this reason, a value obtained by dividing the output P of the D-flip-flop 14 by 1 / m is supplied, for example, as a pixel address when reading an image stored in the pixel memory 15, and an original pixel of a necessary original image is supplied. The remainder of 1 / m of the output P of the D-flip-flop 14 is used as a coefficient processor 17 for performing a filtering process, which determines a ratio of an interpolation process to be performed on pixels read from the pixel memory. Supplied to
[0031]
The pixels read from the image memory 15 are supplied to the filter tap 16, and the output of the filter tap 16 is output as an image output by performing a convolution multiplication with the filter coefficient output from the coefficient processor 7 in the multiplier 18. Note that m is input to the coefficient processor 17 and the up-sampling filter characteristic is selected according to the number of magnifications.
[0032]
When the image is arbitrarily multiplied by 7/5 (1.4 times) with respect to the above operation, m = 7, n = 5, and an image clipping start value = 3 will be described with reference to FIG. explain.
[0033]
At the time of horizontal start, an image cutout start value = 3 is set in the D-flip-flop 14. Thereafter, since the selector 13 is connected to the output side of the adder 2, the count is incremented for each pixel interval base value. Now, since the pixel interval base value = 5, the black triangle mark in the figure represented by the output P of the D-flip-flop 14 is counted with a numerical value as shown in FIG. The value obtained by setting this count value P to 1 / m and the remainder of 1 / m are m = 7, so that the values are as shown in FIG. Since the value of P / m indicates which original pixel the returned pixel corresponds to, the original pixel required for the filter tap is read into the filter tap 16 based on the original pixel specified here. It is possible. Further, the remainder of P / m indicates where the original pixel is located on both sides of the black triangle mark in the figure. This is nothing but a phase when filtering. Therefore, if this value is supplied to the coefficient processor 17, a filter coefficient corresponding to the pixel phase can be selected.
[0034]
In other words, each pixel of the image after conversion is calculated by using this method to determine which of the two original pixels is located in which area, and the value of the required original pixel is interpolated according to the remainder, thereby achieving a comparison with the conventional method. Arbitrary magnification processing can be performed with a very small filter. Therefore, according to the above embodiment, it is possible to perform an arbitrary magnification process while cutting out an image.
[0035]
Next, a second embodiment of the present invention will be described. In the first embodiment, arbitrary magnification is performed by upsampling by m times. However, in order to support m-fold upsampling, m filter characteristics are required. For this reason, the second embodiment is a method in which a multiple of the upsampling is fixed and an arbitrary magnification is made possible by changing the pixel interval base value. With this method, the number of filter characteristics required for upsampling can be reduced.
[0036]
In the present embodiment, a description will be given assuming that the multiple of the fixed upsampling is 16 times.
[0037]
FIG. 3 is a block diagram showing a second embodiment of the system according to the present invention, and FIG. 4 is an operation diagram showing the operation of the second embodiment. Hereinafter, the second embodiment will be described with reference to these drawings.
[0038]
Assuming that m = 7 and n = 5 as in the first embodiment, the magnification is 7/5. Since the upsampling magnification is now 16 times, if the numerator is adjusted to 16 times, the given magnification will be 7/5 = 1/16 / 11.4.
[0039]
Therefore, if the pixel interval base value is set to 11.4, the same as in the first embodiment may be used. However, in this case, the fractional part cannot be realized because it does not exist on the 16-times up-sampled pixel.
[0040]
For this reason, the entire sample interval is made equal by varying the pixel interval base value at certain intervals. For example, in the above example, the pixel interval base value is basically considered to be a numerical value 11, such as 11, 12, 11, 12,... Can achieve a value of 11.4.
[0041]
Therefore, when the magnification of an image is given from the outside of the apparatus in the form of m / n, the calculation means 1 calculates and outputs the reference pixel interval base value as an integer in order to adjust the numerator to 16 times. Next, the calculation means 11 and 21 calculate what period and how many numbers of the pixel interval base values should be corrected so as to correct the integer pixel base values. Thus, the base correction period and the base correction value are obtained.
[0042]
A base correction value is added to the pixel interval base value calculated by the calculation means 11 for each base correction cycle calculated by the calculation means 21. The selector 22 selects a base correction value for each base correction cycle. The output of the selector 22 is supplied to an adder 23 and added to the pixel interval base value. The output of the adder 23 is a pixel interval base value that changes in each base cycle.
[0043]
The output of the adder 23 is supplied to a counter composed of the adder 12 and the D-flip-flop 14.
[0044]
Assuming that the image clipping start value is 7, the output of the D-flip-flop 14 indicates a black triangle mark in FIG.
[0045]
11, 12, 11, 12, 11 and the base value are added to the initial value 7. Now, the pixel interval base value is 11, the base correction period is 1 for each pixel, the base correction value is 1, and the base correction value is added to 11 to be 12.
[0046]
Here, similarly to the first embodiment, 1 / m of the output of the D-flip-flop 14 indicates the pixel position of the original pixel, and the remainder of 1 / m indicates the phase of the filter coefficient for obtaining the interpolation pixel point. In the present embodiment, m is set to 16 times, so that the remainders of 1/16 and 1/16 are bits higher than the fifth bit and lower 4 bits of the D-flip-flop 14, respectively. The 1 / m (9) and the 1 / m remainder (10) required in the first embodiment to be expressed may be merely bit shifts.
[0047]
Thereafter, the procedure for performing the filtering process on the image signal read from the image memory 15 is also calculated in the present embodiment in the same manner as in the first embodiment, and an image output is obtained.
[0048]
According to the above embodiment, an arbitrary magnification process can be performed by a small number of up-sampling filters.
[0049]
In the second embodiment, the upsampling filter has been described as 16 times. However, it goes without saying that the same advantage can be obtained with 32 or 64 as long as it is a power of 2.
[0050]
Next, a third embodiment of the present invention will be described.
[0051]
FIG. 5 is a flowchart showing the third embodiment of the present invention.
[0052]
The third embodiment is a method according to the first embodiment in which a computer system having a microprocessor realizes equivalent processing using software. Hereinafter, the procedure will be described step by step using a flowchart.
[0053]
First, when the magnification of an image is given in the form of m / n, for example, a pixel interval base value is calculated in a pixel interval base value calculation step (S11).
[0054]
Next, the pixel interval base value is supplied to an adding step, and cumulative addition is performed using the pixel interval base value as a base (S12). Then, for the first count (S13), an image cutout start value is added (S14).
[0055]
Then, the cumulative addition output P is reduced by 1 / m to be position information (image-addr) for acquiring an image signal, and the remainder (mod-m) of 1 / m of the cumulative addition output is the phase information of the filter coefficient. (S15).
[0056]
Then, a filter coefficient is selected from mod-m. The coefficients are T tap coefficients in this example.
[0057]
Finally, a multiplication process of the filter coefficient coef [k] (mod-m) and the image signal image [image-addr + k] is performed, and a convolution operation of the filter process is performed, thereby obtaining a converted image output (S16). .
[0058]
Since the operation of this embodiment is the same as that of the first embodiment, the same processing as that of the operation diagram of FIG. 2 is performed by a program using the above-described procedure by a system using a processor.
[0059]
Next, a fourth embodiment of the present invention will be described in detail below.
[0060]
FIG. 6 is a flowchart showing a fourth embodiment of the present invention. The fourth embodiment is a method for realizing the second embodiment by software or the like.
[0061]
First, when the magnification of an image is given in the form of m / n, a pixel interval base value, a base correction value, and a base correction period are calculated in each calculation step so that the numerator is adjusted to 16 times (S21).
[0062]
The base correction period is observed for each calculation cycle for forming the converted image (S22), and when the base correction period is reached, the base correction value is added to the pixel interval base value to correct the pixel interval base (S23).
[0063]
The corrected pixel interval base value is input to the addition step and added (S24). This addition is cumulatively added for each calculation cycle. At the beginning of the image, an image cutout start value is given (S25, S26).
[0064]
In this case, as in the second embodiment, the upsampling magnification is set to 16 times, so that the upper bits of the cumulative addition value are used as position information (image-addr) for acquiring an image signal (S27). The lower bits (mod-m) of the cumulative addition output are used as phase information of the filter coefficient (S27).
[0065]
In the subsequent calculation process, a converted image output is obtained by performing a filtering process in the same manner as in the third embodiment (S28).
[0066]
Since the operation of this embodiment is the same as that of the second embodiment, the image conversion is performed in the same procedure as the operation explanatory diagram of FIG.
[0067]
Although the present invention has been described in detail above, the present invention is not limited to these descriptions, and a similar effect can be obtained as long as a person skilled in the art can assume the same meaning within the scope of the present invention. Needless to say.
[0068]
【The invention's effect】
According to the present invention, for each pixel of the image after the arbitrary magnification conversion, a required original pixel of the original image and a ratio at the time of the filtering processing of the original image are obtained based on the given arbitrary magnification. By performing the arbitrary magnification conversion of the image, the filter circuit necessary for the arbitrary magnification conversion of the image can be greatly reduced.
[Brief description of the drawings]
FIG. 1 is a block diagram showing an example of a first embodiment of a system according to the present invention.
FIG. 2 is an explanatory diagram for explaining the operation of the first embodiment of the system according to the present invention.
FIG. 3 is a block diagram showing an example of a second embodiment of the system according to the present invention.
FIG. 4 is an explanatory diagram for explaining an operation of a second embodiment of the system according to the present invention.
FIG. 5 is a flowchart for explaining the operation of the third embodiment of the system according to the present invention.
FIG. 6 is a flowchart for explaining the operation of the fourth embodiment of the system according to the present invention.
[Explanation of symbols]
11 Calculation means 12 Adder 13 Selector circuit 14 D-flip-flop circuit 15 Image memory 16 Filter tap 17 Coefficient processor 18 Multiplier

Claims (1)

An image arbitrary magnification processing device for enlarging an original image at an arbitrary magnification m / n;
A first calculating unit that calculates a reference pixel interval base value so as to set the numerator of the original image at the arbitrary magnification m / n to “16”;
In order to represent the arbitrary magnification m / n of the original image as “16 / pixel interval specified value”, a base correction value which is a correction value of the pixel interval specified value and a base correction period indicating a period to be corrected are calculated. A second calculator,
A selector for selectively outputting the base correction value calculated by the second calculator according to the base correction cycle;
An adder that receives the pixel interval base value calculated by the first calculator and the base correction value output in the base correction cycle that is the output of the selector, and adds and outputs these;
The counter output P is obtained by receiving the output of the adder and counting by the pixel interval base value or the sum of the pixel interval base value and the base correction value with the image cutout start value of the original image as an initial value. A counter that outputs an upper bit indicating a value of 1/16 of the output P and a lower bit indicating a remainder of 1/16 of the counter output P;
An image memory which receives upper bits indicating a value of 1/16 of the counter output P from the counter, and reads and outputs the original image from a storage area in response thereto;
A coefficient determining unit that receives, from the counter, a lower bit indicating a remainder of 1/16 of the counter output P, and determines a coefficient for performing a filtering process on the original image read from the image memory in response thereto; ,
A multiplier that outputs the original image read from the image memory after performing a filtering process using the coefficient determined by the coefficient determination unit;
An arbitrary image magnification processing device comprising:

Images