JP2007194733A - Image data processor and processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technology of accurately detecting an edge (contour) of an image while excluding the effect of noise, and executing edge processing without making unnaturalness of the image data conspicuous even when the image data include many random noises and fixed pattern noises or digital zoom processing is implemented. <P>SOLUTION: When an average of calculated pixel values indicates a virtual edge and image data of the calculated pixel values include no noise, an image data processing method disclosed herein uses image data of a target pixel for the image data of the target pixel, and in other cases, the image data processing method uses a median value MD for the image data of the target pixel. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image processing technique, and more particularly to a technique for accurately detecting an edge (contour) of an image while eliminating the influence of noise.

In image processing, the edge is detected by various methods (for example, Patent Documents 1 to 4).
In many cases, noise is superimposed on image data. For example, when taking an image of a dark place at night, it is necessary to increase the gain of an imaging means, for example, a CCD or CMOS sensor, and noise is increased. Also, when data zoom processing is performed, noise may be magnified and the image may deteriorate. Furthermore, electrical noise may be superimposed on the imaging signal from the inside of the imaging device or from the outside of the imaging device. In addition, when an image pickup signal from a CCD, CMOS sensor or the like is converted to digital, it may be affected by errors due to variations in conversion of the analog / digital converter.
Therefore, it is necessary to eliminate the influence of noise in edge detection.

  As a countermeasure against such noise processing, it has been proposed to use a median filter. However, when the median filter is applied to the entire image data, the edge of the image is blurred and the resolution is lowered.

Patent Document 5 proposes to overcome such a problem. An outline of the contents described in Patent Document 5 will be described. Values from the maximum value to the nth value of the image data of the pixel of interest (pixel of interest) and the image data around the pixel of interest are obtained. Further, the values from the minimum value to the m-th value of the image data of the pixel of interest and the image data around the pixel of interest are obtained. When the difference between the maximum value and the minimum value is larger than the threshold value, it is determined as an edge, and the image data of the target pixel is used as the image data of the target pixel. On the other hand, when the difference between the maximum value and the minimum value is smaller than the threshold value, it is determined that it is not an edge, and the image data of the pixel of interest is set as the median value.
JP 2005-149266 A JP 2005-275681 A JP 2004-159331 A JP 2004-272885 A JP 2004-16601 A

However, the method described in Patent Document 5 encounters the problems described below.
Since the median filter of the pixel of interest is turned on / off only with the image data of 9 pixels, the order of the processing results of the image data of the pixels in the flat portion is random in the image data with a lot of random noise and fixed pattern. , “Median value”, “Median value”, “Original image data”, “Median value”, and so on. May stand out.
In particular, when digital zoom processing is performed, the unnaturalness of the image is noticeable.

  Further, various edge processing techniques have been proposed in Patent Documents 1 to 4, but there is a demand to detect edges quickly and accurately with a simple method or with a simple device configuration.

According to the present invention, for each of a plurality of pixels arranged two-dimensionally, calculating an average value and a median value for surrounding pixels centered on each pixel;
Determining whether the calculated average value of pixels is a virtual edge;
Determining whether the calculated pixel image data includes noise;
If the result of the determination is a virtual edge and no noise is included, the step of setting the image data of the pixel of interest as the image data of the pixel of interest;
If the result of the above judgment is a virtual edge that contains noise, if it is not a virtual edge and does not contain noise, or if it contains noise instead of a virtual edge And an image data processing method comprising: setting a corresponding median value as image data of a pixel of interest.

According to the present invention, there is provided an apparatus for performing the image data processing method.
That is, according to the present invention, for each pixel of a plurality of pixels arranged two-dimensionally, means for calculating an average value and a median value for surrounding pixels centered on each pixel;
Means for determining whether or not the calculated average value of pixels is a virtual edge;
Means for determining whether noise is included in the calculated pixel image data;
If the result of the determination is a virtual edge and noise is not included, means for setting the image data of the pixel of interest as the image data of the pixel of interest;
If the result of the above judgment is a virtual edge that contains noise, if it is not a virtual edge and does not contain noise, or if it contains noise instead of a virtual edge And an image data processing device including means for setting the corresponding median value as image data of the pixel of interest.

The present invention performs final edge determination processing based on the results of virtual edge determination and noise determination. If it is a virtual edge and no noise is included, the image data of the target pixel is set as the image data of the target pixel. Otherwise, the corresponding median value is set to the target pixel P _ij . Since the image data is used, noise reduction can be performed while preserving the edge of the image even when there is a lot of random noise or fixed pattern noise.

The image data processing method of the present invention is simple and can be processed quickly.
Also, the configuration of the image data processing apparatus of the present invention is simple and can be processed quickly.

Embodiments of a method for detecting an edge of an image and an apparatus for detecting an edge of an image according to the present invention will be described with reference to the accompanying drawings.
FIG. 1 is a configuration diagram of an image processing apparatus including a portion for detecting an edge of an image according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating an outline of an arrangement of image data for one frame.
FIG. 3 is a flowchart for explaining image processing, particularly edge processing, in the apparatus for detecting the edge of the image illustrated in FIG.

Configuration of Image Processing Device An image processing device 1 according to an embodiment of the present invention includes an imaging unit 10, a signal processing unit 20, a frame memory 30, and a display device 50.
As the image pickup means 10, various image pickup means such as a CCD (Charge Coupled Device), a C-MOS sensor and the like can be used. Hereinafter, a case where the CCD 100 is used as the imaging unit 10 will be described.
The signal processing means 20 is means for performing image signal processing described below. For example, a signal processing processor such as a computer can be used. Hereinafter, the case where a signal processor is used as the signal processing means 20 will be exemplified.

As illustrated in FIG. 1, the signal processor 200 as an example of the signal processing unit 20 includes a central processing unit (CPU) 201, a ROM 202 and a RAM 203 in which programs for performing processing to be described later, fixed data, and the like are stored. , An interface (I / F) unit 204 with the CCD 100, an analog / digital (A / D) conversion unit 205, a work memory 206, an interface (I / F) unit 207 with the display device 50, and a bus 208.
These ROM 202, RAM 203, I / F unit 204, A / D conversion unit 205, work memory 206, and I / F unit 207 are connected to the CPU 201 via a bus 208.
The RAM 203 is used for temporarily storing data in signal processing to be described later.

  The CPU 201, the signal processor 200 including the work memory 206, and the frame memory 30 constitute each unit of the image data processing apparatus of the present invention.

In this embodiment, the work memory 206 stores the average image value for each pixel, stores the presence / absence of virtual edges, stores the presence / absence of noise, and the presence / absence of final edges. Used for memory.
For example, when the logic “1” is set in the memory portion of the work memory 206 at the pixel position indicating the virtual edge determined to be a virtual edge, and it is determined that the edge is not a virtual edge Is set to logic "0".
Similarly, when it is determined that the image data of the pixel of interest P _ij includes noise, a logic “1” is set in the corresponding portion of the work memory 206 at the pixel position, and noise is included. When it is determined that there is no logic, logic “0” is set.
When the memory portion of the work memory 206 at the pixel position indicating the edge detection where it is determined that the edge is finally detected is set to logic “1”, and it is finally determined that there is no edge. Is set to logic "0".
Thus, looking at the work memory 206, the state of each pixel is, for example, whether there is a virtual edge, whether noise is included, or whether an edge is finally detected. Can be listed.
Although the case where the RAM 203 and the work memory 206 are provided independently has been described in the present embodiment, the RAM 203 can be used as the work memory 206.

The frame memory 30 is a memory that stores an image captured by the CCD 100 for one frame. Note that, as illustrated in FIG. 2, one frame is assumed to be composed of (m columns in the horizontal direction) × (n rows in the vertical direction) pixels.
In the present embodiment, a case will be described in which the frame memory 30 stores three types of signals for each pixel: red (R), green (G), and blue (B). In the following description, a case where edge processing is performed for each of red (R), green (G), and blue (B) read from the CCD 100 will be described.
Of course, the frame memory 30 can store luminance data and two types of color difference data instead of the red (R), green (G), and blue (B) signals. In that case, the edge processing may be performed only for the luminance signal.

  The display device 50 is a display device that displays the processing result of the signal processor 200, for example, an LCD.

Outline of processing contents of image processing apparatus The operation of the image processing apparatus 1 will be described with reference to FIG.

Step 1: The imaging processing CPU 201 drives the CCD 100 via the I / F unit 204.
The CPU 201 inputs the imaging result of the CCD 100 via the I / F unit 204, converts an analog imaging signal into a digital value by the A / D conversion unit 205, and sequentially converts the converted digital imaging signal to the frame memory 30. I will remember it.
The frame memory 30 stores image data for one frame, for example, three types of image data of RGB components in the frame memory 30.
For example, as illustrated in FIG. 2, the image data stored in the frame memory 30 is one-dimensionally arranged in a two-dimensional manner (matrix) in the horizontal direction and the vertical direction. One frame of image data is composed of m pixels in the horizontal direction and n pixels in the vertical direction for each of the three types of image data of the RGB components.

Step 2, the display processing CPU 201 reads the image data stored in the frame memory 30 according to the display method of the display device 50 and outputs it to the display device 50.
As a result, the image captured by the imaging unit 10 is displayed on the display device 50.

Step 3 Check for Edge Determination The CPU 201 checks whether or not it is time to perform edge determination processing. If edge determination is necessary, the CPU 201 performs edge determination processing.
The edge determination process can be performed every time imaging is performed, for example, 30 times per second, or can be performed intermittently at a predetermined cycle, for example, once per second.
Alternatively, it can be performed on demand when there is an instruction or request from a user or another device.

Step 4, Edge Determination Processing When the edge determination processing is necessary, the CPU 201 performs edge determination processing described with reference to FIG. 4 (first embodiment) and FIG. 7 (second embodiment).

Edge determination processing, first embodiment The edge determination processing of the first embodiment will be described with reference to FIG.

Step 11: Average Image Data Value Calculation and Median Value Calculation The CPU 201 calculates the average image data value AV and the median value MD as preprocessing of the edge determination process.
The image data stored in the frame memory 30 is, for example, as illustrated in FIG. 2, each of the RGB component image data for one frame is m pixels in the horizontal direction and n pixels in the vertical direction. It consists of The R component image data will be described below as a representative.

Calculation of average image data value AV The average image data value AV is used to determine a virtual edge.
In the present embodiment, as illustrated in FIG. 5A, the image data P _i− of a total of 9 pixels, with one pixel around the pixel of interest (specific pixel or specific pixel) P _ij as the center. _{1, j} total for _{~P i + 1, j + 1} , obtaining an average image data values AVP _ij of the target pixel P _ij is divided by summing the number = 9.
This is calculated for the image data of all pixels for one frame.
Specifically, the CPU 201 reads out the image data of the pixel of interest P _ij and the surrounding eight pixels, for example, the image data of the R component from the frame memory 30, stores it in the RAM 203, and stores these nine pixels. The images are added and divided by 9 to obtain the average value AVP _ij of the pixel of interest P _ij . The CPU 201 stores the obtained average value AVP _ij in the average image data value storage portion of the corresponding pixel P _{ij in} the work memory 206.

As a method for calculating the average image data value AVP _ij , in addition to the above-described simple average value, an average value can be obtained in consideration of a weighting factor. For example, the weighting factor of the pixel of interest P _ij is 2, the weighting factor of the surrounding pixels is 1, the image data of each pixel multiplied by the weighting factor is summed, and the addition result is divided by 10 and weighted An average value AVP _ij can also be obtained. The CPU 201 stores the weighted average value AVP _ij thus obtained in the average image data value storage portion of the target pixel P _{ij in} the work memory 206.

In the vicinity of one frame, there is no pixel data for one pixel around the pixel of interest.
Therefore, the CPU 201 uses, for example, the average value of the pixel of interest at the four corners, for example, for the pixel of interest at the left corner, as illustrated in FIG. 5B. Ask for. As described above, whether to obtain a simple average value or a weighted average value is appropriately determined.
The CPU 201 stores the average image data average value AVP _ij obtained in this way in the average image data value storage portion of the pixel of interest P _{ij in} the work memory 206.
Since there are no edges outside the pixel of interest at the four corners, the CPU 201 does not perform edge determination processing outside the pixel of interest at the four corners.

Similarly, with respect to the pixel of interest in the peripheral portion of the frame memory 30, for example, as illustrated in FIG. 5C, for example, with respect to the pixel on the upper side, the average image with the image data of the surrounding five pixels Obtain the data average.
The CPU 201 stores the average image data value AVP _ij thus obtained in the average image data value storage portion of the pixel of interest P _{ij in} the work memory 206.
Also in this case, since there is no edge outside the side, the CPU 201 does not perform the edge determination processing outside the pixels of these sides.

Median value calculation The median value MD is used when determining the final value of the pixel of interest.
In parallel with the calculation of the average image data value AV, the CPU 201 obtains a median value MD, that is, a median value.
The method for obtaining the median value MD is the same as the median value calculation described in Patent Documents 1 to 5 described above, for example.
The CPU 201 stores the calculated median value MD in the median value storage area of each pixel P _{ij in} the work memory 206.

Steps 12-15: Virtual Edge Determination Processing Once the average image data value is obtained for each pixel, the CPU 201 determines that the average image data value for each pixel of interest is the edge in relation to the image data of adjacent pixels. It is virtually determined whether or not there is.
The virtual edge means that it is temporarily determined as an edge in this process with respect to the final edge determination described later.

Step 12, the edge calculation process in the first embodiment of the difference calculation is performed for the average image data value of the pixel of interest and the average image data value of surrounding pixels centered on the pixel of interest arranged in the form of a macro risk. Find the difference between the maximum and minimum average image data values in the left half pixel area in the diagonal direction of the pixel, and the difference between the maximum and minimum average image data values in the right half pixel area. To do that.
FIG. _6A is a diagram illustrating the image data of the pixel of interest P _ij , for example, the average image data value AVP _{ij of} the R component and the average image data value AV of the surrounding pixels.
The CPU 201 arranges the average image data values AV arranged in a two-dimensional shape illustrated in FIG. 6A in one column as illustrated in FIG. 6B and stores the average image data values AV in, for example, the RAM 203.
The CPU 201 obtains a difference Lδ on the left side and a difference Rδ on the right side based on the average image data value AVP _ij of the pixel of interest P _ij .
That is, the CPU 201 uses the maximum value of the five average image data values AVP _{i−1, j−1} (A ₀ ) to AVP _{i, j} (A ₄ ) in FIG. MAX) and the difference between the minimum value (MIN).
Similarly, the CPU 201 sets the maximum value of the five average image data values AVP _{i, j} (A ₄ ) to AVP _{i + 1, j + 1} (A ₈ ) in FIG. The difference between (MAX) and the minimum value (MIN) is obtained.

If the data array in FIG. 6A is changed to the data array as shown in FIG. 6B, the RAM 203 is used to determine the left difference Lδ and the right difference Rδ for the next pixel of interest P _{i + 1, j.} In FIG. 6B, the data of A _{0 to} A ₈ in FIG. 6B are shifted to the left one by one, A ₀ is erased, and image data of the next pixel of AVP _{i + 1, j + 1} (A ₈ ) Since the above-described processing can be easily performed by inserting A at the position of A ₈ , signal processing is simplified.
Furthermore, since only nine RAMs 203 are used for the calculation of the left difference Lδ and the right difference Rδ, there is an advantage in terms of reduction in memory usage.

Steps 13-15: The virtual edge determination CPU 201 determines that the pixel of interest P _ij is virtual when the left-side difference Lδ is greater than or equal to a predetermined threshold or when the right-side difference Rδ is greater than or equal to a predetermined threshold. Assuming that the edge portion is an edge portion, the virtual edge flag EFLG = 1 is set.
On the other hand, when the left-side difference Lδ is equal to or smaller than the predetermined threshold value and the right-side difference Rδ is equal to or smaller than the predetermined threshold value, the CPU 201 assumes that the pixel of interest P _ij is not virtually an edge portion, The virtual edge flag EFLG = 0 is set.

Steps 16 to 18: Determination of whether or not noise is included The CPU 201 obtains a deviation between the image data of the pixel of interest P _{ij and} the image data of surrounding pixels illustrated in FIG.
Since the image data of the pixel of interest P _ij and the surrounding image data are generally continuous, it may be assumed that they are within a certain degree of deviation except when a large amount of noise is included instantaneously.
Therefore, the CPU 201 determines that noise is included in the image data of the pixel of interest P _ij when the absolute value of the deviation exceeds a predetermined level. When it is determined that noise is included, the CPU 201 sets the noise flag NFLG = 1, and when it determines that noise is not included, the CPU 201 sets the noise flag NFLG = 0.

Steps 19 to 25, the final edge determination CPU 201 determines that the virtual edge flag EFLG = 1, that is, the image data of the pixel of interest P _ij is a virtual edge, and the noise flag NFLG = 0, that is, the pixel of interest P _ij If the image data does not contain noise, the image data of the pixel of interest P _ij is used as it is, assuming that it is finally an edge.
As described above, when it is determined that an edge is not included in noise, the original image data of the pixel of interest P _ij is used, so that the edge is not blurred. In particular, even when there is a lot of random noise or fixed pattern noise, it is possible to perform noise reduction while preserving image edge data.
In all other cases, the median value MD stored in the work memory 206 is used as the image data of the pixel of interest P _ij .
The above processing is performed for all the image data illustrated in FIG.

The modification mode CPU 201 of the first embodiment can store the value of the virtual edge flag EFLG described above in a memory portion indicating the virtual edge of the pixel of interest P _{ij in} the work memory 206. Further, the CPU 201 can store the value of the noise flag NFLG in the memory portion corresponding to the pixel of interest P _{ij in} the work memory 206. As a result, when the work memory 206 is observed for one frame of image data, a list of virtual edge states and whether or not noise is included can be listed.

  Further, when the pixels of the virtual edge flag EFLG = 1 and the noise flag NFLG = 0 stored in the work memory 206 are displayed on the display device 50, for example, which portion of the image data of one frame is finally used as an edge It is possible to list whether the determination has been made.

More preferably, the CPU 201 checks the continuity of the final edge result in the work memory 206.
Usually, it does not protrude by one pixel and is not an edge. Edges are usually continuous across multiple pixels. Therefore, the CPU 201 checks the continuity of the result of the last edge in the work memory 206, and stores only a portion determined as an edge over a plurality of pixels in the work memory 206 as a formal edge.

Although the above-described processing has been described for the case where only the R component is performed among the RGB components, for example, the above-described processing can also be performed for each of the remaining GB components.
Alternatively, in order to simplify the processing, when it is only necessary to perform processing on the main component of the video, for example, the G component, the above-described processing can be performed only on such a component.

  Of course, in the embodiment of the present invention, the above-described processing can be performed only for the luminance signal component, not for the RGB component, but for the luminance component and the two color difference component signals.

Edge determination process, second embodiment In the first embodiment of the edge determination process described with reference to FIG. 4, the case where the edge determination is performed once is described, but the image illustrated in FIG. According to the data arrangement, the edge determination can be performed separately in the horizontal direction and the vertical direction.
Such an example will be described with reference to FIG. 7 regarding the edge determination processing according to the second embodiment of the present invention.

Step 31, average image data value calculation and median value calculation The CPU 201 calculates an average image data value AV and a median value MD as preprocessing of edge processing.
The calculation process of the average image data value AV and the calculation process of the median value MD are the same as the process of step 11 described with reference to FIG.

Steps 32 to 34, whether or not noise is included The CPU 201 determines the image data of the pixel of interest P _ij illustrated in FIG. 5A as described in steps 16 to 18 of the first embodiment. And the deviation from the image data of the surrounding pixels.
Since the image data of the pixel of interest P _{ij and} the image data of the surrounding pixels are generally continuous, it is assumed that they are within a certain degree of deviation except when excessive noise is instantaneously superimposed on these image data. You can do it.
When the absolute value of the deviation exceeds a predetermined level, the CPU 201 determines that the image data of the pixel of interest P _ij includes noise.
If it is determined that noise is included, the CPU 201 sets the noise flag NFLG = 1, and if it determines that noise is not included, the CPU 201 sets the noise flag NFLG = 0 and further sets the image data of the pixel of interest P _ij to the median value MD. And

As illustrated in the flowchart of FIG. 4, only when the virtual edge flag EFLG = 1 and the noise flag NFLG = 0, the raw image data of the pixel of interest P _{ij is} used as the image data of the pixel of interest P _ij as the final edge. Used.
Therefore, also in the second embodiment, the following processing is performed only when the noise flag NFLG = 0, and when the noise flag NFLG = 1, the image data of the pixel of interest P _ij is replaced with the value of the median value MD, The following processing is not performed.

In steps 35 to 38, the horizontal virtual edge determination CPU 201 obtains a deviation from the average image data before and after the horizontal direction of the pixel of interest P _ij illustrated in FIG.
δ1 _H = AVP _{i−1, J to} AVP _{i, J} ,
δ2 _H = AVP _{i, J to} AVP _{i + 1, J}

When both δ1 _H and δ2 _H are within a predetermined determination level, it is determined that the pixel of interest P _ij has no edge in the horizontal direction, and the CPU 201 sets the horizontal virtual edge flag EFLG _H = 0. In this case, CPU 201 sets the image data of the target pixel P _ij and the image data of the target pixel P _ij.
On the other hand, when either δ1 _H or δ2 _H exceeds a predetermined determination level, it is determined that the pixel of interest P _ij has an edge in the vertical direction, and the CPU 201 sets the virtual edge flag EFLG _{H in the} vertical direction = Set to 1. In this case, the CPU 201 sets the image data of the pixel of interest P _ij as the corresponding median value MD.

Step 39-43, determination CPU201 vertical virtual edge, at step 39, a deviation of the vertical direction of the front and rear of the average image data of the target pixel P _ij that illustrated in FIG. 6 (A).
δ1 _V = AVP _{i, J-1 to} AVP _{i, J} ,
δ2 _V = AVP _{i, J to} AVP _{i , J + 1}

When both δ1 _V and δ2 _V are within a predetermined determination level, it is determined that the pixel of interest P _ij has no edge in the vertical direction, and the CPU 201 sets the virtual edge flag EFLG _V = 0 in the vertical direction in step 41.
Further, when it is determined that there is an edge in the horizontal direction when the horizontal virtual edge flag EFLG _H = 1, the CPU 201 stores the image data of the pixel of interest P _ij as it is as the image data of the pixel. .
On the other hand, when the virtual edge flag EFLG _H in the horizontal direction is 0, it is determined that there is no edge in both the horizontal direction and the vertical direction, so the image data of the pixel of interest P _ij is set as the median value MD.

When either δ1 _V or δ2 _V exceeds the predetermined determination level, it is determined that the pixel of interest P _ij has an edge in the vertical direction, and the CPU 201 sets the virtual edge flag EFLG _V = 1 in the vertical direction. . Further, the CPU 201 stores the image data of the pixel of interest P _{ij as it is as} the image data of the pixel.

The above processing is performed for all the image data illustrated in FIG.
As described above, it is determined whether or not there is an edge in the horizontal direction and the vertical direction of the pixel of interest _Pij .
The CPU 201 can store the horizontal virtual edge flag EFLG _H and the vertical virtual edge flag EFLG _V in corresponding portions in the work memory 206, respectively. Further, as described above, the corresponding image data storage portion of the work memory 206 stores the image data or median value MD of the pixel of interest P _ij .

Also in the second embodiment of the edge determination process, when it is determined that the edge is in the horizontal direction and / or the vertical direction without noise, the original image data of the pixel of interest P _ij is used, so that the edge is There is no blur. In particular, even when there is a lot of random noise or fixed pattern noise, it is possible to perform noise reduction while preserving image edge data.
In all other cases, the median value MD stored in the work memory 206 is used as the image data of the pixel of interest P _ij . Thereby, noise is reduced.
In the second embodiment, in particular, an edge in consideration of the directionality in the horizontal direction and / or the vertical direction can be detected.

If you want to determine the edges of the variant further oblique direction, CPU 201 can, in FIG. 6 (A), the of the front and rear diagonal direction of the target pixel P _ij pixel and the average value image data, horizontal and vertical as described above Processing similar to the processing can be performed.

  Also in the second embodiment, either luminance signals or RGB can be used as image data to be subjected to noise determination processing and edge determination processing.

The implementation of the image data processing method and apparatus of the present invention is not limited to the above-described examples, and various modifications can be made.
For example, in the noise processing with reference to FIG. 5 (A), the example in which the deviation of the image data is calculated for one pixel of the outer periphery of the pixel of interest P _ij has been described. It can also be expanded to minutes. Of course, it is not limited to 2 pixels perimeter.
Similarly, in the virtual edge determination process described with reference to FIGS. 6A and 6B, the average image data value of nine pixels centered on the average image data value AV of the pixel of interest P _ij. Although the case where AV is used has been exemplified, virtual edge determination can also be performed with reference to the average image data value AV for two outer peripheral pixels. Of course, it is not limited to 2 pixels perimeter.

From the above description, the signal processor 200 (signal processing means 20) illustrated in FIG. 1, in particular, the CPU 201, the work memory 206, and the frame memory 30 constitute various means of the image data processing apparatus of the present invention. It will be understood that
Similarly, the operation processing of the signal processor 200 (signal processing means 20) illustrated in FIG. 1, in particular, the CPU 201, the work memory 206, and the frame memory 30 is processed in various steps of the image data processing method of the present invention. It will be understood that this is implemented.

FIG. 1 is a configuration diagram of an image processing apparatus including a portion for detecting an edge of an image according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an outline of an arrangement of image data for one frame. FIG. 3 is a flowchart for explaining image processing, particularly edge processing, in the apparatus for detecting the edge of the image illustrated in FIG. FIG. 4 is a flowchart of the first embodiment of the edge determination method. FIGS. 5A to 5C are diagrams for describing an average value of image data as preprocessing for edge determination. 6A and 6B are diagrams illustrating the image data of the target pixel and the average image data value of the image data of the surrounding pixels. FIG. 7 is a flowchart of the second embodiment of the edge determination method.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Image processing apparatus 10 ... Imaging means,
20 ... Signal processing means,
201 ... CPU, 202 ... ROM, 203 ... RAM, 204 ... I / F unit,
205 ... A / D conversion unit, 206 ... work memory, 207 ... I / F unit,
208 ... Bus
30: Frame memory,
50. Display device

Claims (6)

For each pixel of a plurality of pixels arranged in a two-dimensional manner, calculating an average value and a median value for surrounding pixels centered on each pixel;
Determining whether the calculated average value of pixels is a virtual edge;
Determining whether the calculated pixel image data includes noise;
If the result of the determination is a virtual edge and no noise is included, the step of setting the image data of the pixel of interest as the image data of the pixel of interest;
If the result of the above judgment is a virtual edge that contains noise, if it is not a virtual edge and does not contain noise, or if it contains noise instead of a virtual edge An image data processing method comprising: setting a corresponding median value as image data of a pixel of interest. In the step of determining the virtual edge,
With respect to the average image data value of the pixel of interest and the average image data value of surrounding pixels centering on the pixel of interest, the average image of the pixel region in the substantially left half in the diagonal direction of these pixels arranged in the form of a macro risk Find the difference between the maximum and minimum data values, and the difference between the maximum and minimum average image data values in the pixel area in the right half of the right,
When one of the differences between the two exceeds a predetermined threshold, it is determined as a virtual edge.
The image data processing method according to claim 1. In the step of determining the virtual edge,
Regarding the average image data value of the pixel of interest and the average image data value of surrounding pixels centered on the pixel of interest,
If the difference between the average image data value of the pixel of interest and the average image data value of the pixels before and after the pixel of interest in the horizontal direction of these pixels arranged in a macrosquist, exceeds a predetermined threshold value, Judge that there is an edge in the horizontal direction,
When the difference between the average image data value of the pixel of interest and the average image data value of the pixels before and after the pixel of interest in the vertical direction of these pixels arranged in the form of a macro risk exceeds a predetermined threshold value, Determine that there is an edge in the vertical direction,
The image data processing method according to claim 1. In the step of determining the virtual edge,
Regarding the average image data value of the pixel of interest and the average image data value of surrounding pixels centered on the pixel of interest,
The difference between the average image data value of the pixel of interest and the average image data values of the pixels before and after the pixel of interest exceeds a predetermined threshold in the first diagonal direction of these pixels arranged in the form of a macro risk. It is determined that there is an edge in the first diagonal direction,
The difference between the average image data value of the pixel of interest and the average image data values of the pixels before and after the pixel of interest in a second diagonal direction of these pixels arranged in a macroslice exceeds a predetermined threshold value. If it is determined that there is an edge in the second diagonal direction,
The image data processing method according to claim 3. In the noise determination step,
When the difference between the image data of the pixel of interest and the image data around the pixel of interest exceeds a predetermined level, it is assumed that the image data of the pixel of interest contains noise.
The image data processing method according to claim 1. Means for calculating an average value and a median value for surrounding pixels centered on each pixel, for each of a plurality of pixels arranged in a two-dimensional manner;
Means for determining whether or not the calculated average value of pixels is a virtual edge;
Means for determining whether noise is included in the calculated pixel image data;
If the result of the determination is a virtual edge and noise is not included, means for setting the image data of the pixel of interest as the image data of the pixel of interest;
If the result of the above judgment is a virtual edge that contains noise, if it is not a virtual edge and does not contain noise, or if it contains noise instead of a virtual edge And a means for setting the corresponding median value as image data of the pixel of interest.

Images