JPH04227181A - Edge detector - Google Patents

Abstract

PURPOSE: To provide an edge section detector which can excellently detect the direction of the edge section of a picture area containing highly detailed vertical frequency sections by improving the performance of the already existing detector and providing such a function that increases the selectable number by means of a direction detecting means. CONSTITUTION: First means 40-44d generate first signals d1 -d5 which change correspondingly to the values of primary pixels A, B, and C on a prescribed line in a present field and related secondary pixels 1, 2, and 3 on the adjacent line. An evaluating means 46 generates an edge section direction signal based on the first signals d1 -d5 and processes the secondary pixels 1, 2, and 3 to those which are contained in the succeeding field of the present field and positioned on the adjacent line of the prescribed line in the present filed after compensation for motion. Therefore, the edge section direction detecting ability of an edge section detector is further improved.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention This invention relates to video equipment, and more particularly to edge direction detectors for video equipment.

0002

BACKGROUND OF THE INVENTION Interlaced display standards increase artifacts or line crawwells.
e-crawl) and interline flicker.
It is known that this can be significantly reduced or eliminated by converting to a progressive scan scheme. To perform this sequential scanning, lines are inserted by line repetition or interpolation between each pair of adjacent lines in each field of the interlaced scanning system. A cost effective way to perform interpolation is to use intermediate filters. Intermediate filters work effectively on step response and noisy images. However, since some known algorithms use only vertical interpolation, the use of intermediate filters results in perturbed serration effects along the edges of moving diagonal transitions (
(serration effect) will occur. The visual intensity of this serration effect depends on, for example, the contrast and display characteristics of the image, and is generally clearly visible to the naked eye on a 20-inch television receiver or monitor. Since the display size of larger televisions is even larger, the sales effect described above becomes particularly pronounced.

European Association for Signal Processing IEEE, EI
Sponsored by C, ITE and SMPTE 198
Proceedings of the 9th year held in Turin
The Workshop on HDTV (Proc
eedings of the workshop o
n HDTV)
Scan Conversion Using Edge
Information (Progressive Sc
An Conversion Using Edge
Information)”, T. Doyle (T.
Doyle) and M. Looy
It was proposed in J. mans) to reduce the above-mentioned effects by detecting the direction in which the edges extend and changing the interpolation algorithm based on the detected direction. In this document, a 2x3 window of the current field is used, the magnitude of three pixel differences in three different directions is determined, and the smallest of these values indicates the direction of the edge. I'm assuming. For the assumed vertical edge, an intermediate filter is operated on two selected pixels from the current field, namely the pixel where the above-mentioned minimum difference occurred and the corresponding pixel from the previous field. . For the other directions, the average value of the pixel pair in which the smallest absolute value of the difference occurs is used. To provide direction selection that is independent of high frequency noise, a low pass filter is used to filter the input video data.

0004

Although the known edge direction detectors described above can significantly reduce artifacts, they still have drawbacks with high frequency images, and further improvements are desired. It is therefore an object of the present invention to further improve the performance of the edge detector described above, in particular for image areas containing high frequency details.

Another object of the present invention is to provide an edge detection device capable of detecting additional edge directions near the horizontal axis. It is particularly effective because the main artifact of the intermediate filter, the serration effect on moving slope transitions, becomes more pronounced in the horizontal direction.

[0006]

SUMMARY OF THE INVENTION The present invention provides an apparatus for approximating the direction of the edges of a video image formed by a video signal consisting of sequential fields and sequential lines within each sequential field. It is in. The device according to the invention comprises a first signal generating a first signal representative of the difference between the pixel value of each of the primary pixel groups in a line of the current field and the pixel value of the respective corresponding slope defining the secondary pixel. and means for generating an edge direction signal as a function of the first signal. And, the present invention provides the above-mentioned 2
The slope defining the next pixel is characterized in that the slope defining the next pixel is a pixel in a vertically adjacent line in a field subsequent to the current field.

The operation and configuration of the present invention will be explained in detail below based on the drawings.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In order to reduce the artifacts that are noticeable with interlaced scanning, namely line crawlers and interline flicker, it has been proposed to convert to a progressive scanning scheme. To perform this progressive scanning scheme, lines are inserted between each pair of adjacent lines of each field of the interlaced scan by line repetition or interpolation, preferably by intermediate filters. However, since only vertical interpolation is used in some known algorithms, the use of intermediate filters results in a serration effect that is perturbed along the edges of the moving diagonal transition.

[0009] In the above-mentioned literature, it has been proposed to detect the orientation of edges and to change the interpolation algorithm based on the detected edges. In this method, a 2x3 window of the current field is used and the difference A-
C', C-A', and B-B' are determined and the minimum of these values is assumed to indicate the edge direction. For the assumed vertical edge, an intermediate filter is operated on two selected pixels from the current field, namely the pixel for which the above-mentioned minimum value was obtained and the corresponding pixel D from the previous field. There is. For other directions,
The average value of the pixel pair for which the smallest absolute value of difference occurs is used. FIG. 2 is a block diagram showing the configuration of the known device described in the above-mentioned literature. As shown in FIG. 2, a low pass filter 26 filters the input video data.
is added to perform direction selection independent of the high-frequency nozzle. The input video data is provided to an interpolator 24 whose interpolation direction is determined by an edge direction detector stage 22. The low pass filter 26 mentioned above filters the video data before providing it to the detection stage 22. Finally, the video data and the output of interpolator 24 are multiplexed by multiplexer 28.

As stated above, it is an object of the present invention to improve the performance of known detectors for image areas containing high vertical frequency detail and to provide the ability to increase the number of directions that can be selected by the direction detector stage. It's about achieving. The shorter the distance between the samples for determining the difference, the more reliable the direction determination in the high frequency region, and the final selection of the slope based on the difference. As shown in FIG. 1, in conventional apparatus the values used to determine the difference or slope in each of the three directions are vertically spaced two frame lines apart. On the other hand, in the present invention (FIG. 3), samples A, B, and C (herein also referred to as primary pixels) of the current field are similarly arranged as samples A', B', C' ( (also referred to as secondary pixels), but instead places the line vertically midway between the line with samples A, B, C and the line with samples A', B', C'. Process with samples 1, 2, and 3 from the next field to be generated. Considering the difference shown on the right side of FIG. 3, the directions d1 and d5 are
The most horizontally inclined differences A-C' and C-A' in FIG.
It is in a direction that is even closer horizontally than the Direction d2, which is based on the absolute value of the difference between pixel A and pixel 2, produces the same slope as direction A-C' in FIG. 1, but with a length 1/2. Therefore, the accuracy in this direction is improved by a factor of two. Similarly, the absolute value difference in the direction d3 is the absolute value difference BB′ in FIG.
This is 1/2 the size of , which also improves the vertical direction. The difference between directions d4 and d5 is also the same as the difference between directions d2 and d1. Therefore, a closer orientation in the horizontal direction is obtained, and the distance based on the difference between the samples is also shorter than in the prior art.

FIG. 4 shows the configuration of an example of an edge detector according to the present invention. The input video signal is delayed in successively connected pixel delay elements 42 and 42a, which output pixels A and B, respectively, to their outputs. The input video signal is also delayed in field delay element 40 and further pixel delay elements 42b and 42c. As shown in FIG. 4, pixel 3 is output from field delay element 40. Pixels 2 and 1 are pixel delay elements 4
2b and 42c, respectively. The pixel values extracted from this delay circuit are supplied to subtraction stages 44, 44a, . . . according to the table of FIG. A subtraction stage 44 or the like calculates the absolute value of the difference between the signals applied to these inputs. Calculated difference d1, d2, −
--- is fed to a comparator circuit 46 and an edge direction evaluation is performed. This output will be the direction associated with the minimum absolute value of the difference, or the result of ranking the differences by absolute value, with intermediate differences in the direction intermediate between the directions represented by the maximum and minimum difference values. Minimum difference should be used only when relevant.

It has been found useful to filter the vertical video signal before calculating the edge direction to avoid artifacts due to moving parts of the image. coefficient 1
Good results are obtained using a 2-tap transversal filter with /2 and 1/2. Coefficient 1/4
, 1/2, and 1/4 provide better results.

Performing motion correction on the values from the previous field before providing the signal to the edge detector further improves the overall operation of the edge detector. In a second preferred embodiment using samples from the previous field, an arctangent or arccotangent function is used to determine the directional angle. In the simplest form of this embodiment, a look-up table is used to find the angle, for example, tangent (A-B)/(A-1) using the pixels shown in FIG. Since the look-up table is suitably quantized in any way, i.e. the total number of directional angles to be calculated can be retrieved, its distribution can increase the resolution of a certain angular region, and this method is particularly flexible. It is excellent. In order to evaluate each edge direction above,
Only three pixels are required. Further improvements can be made using a larger number of pixel values. As shown empirically, the ratio

[0014]

[Outside 1]

##EQU1## approximates the tangent of the intersection angle of the edges assumed in the horizontal direction more accurately than the tangent approximation described above. This intersection angle is

[0016]

[Outside 2]

It can be shown mathematically that . In the embodiments described above, the edge detector according to the present invention was explained using a scanning rate conversion device, but the present invention is not limited to this device, but can be applied to other devices, and can be modified in various ways. is possible.

[Brief explanation of the drawing]

FIG. 1 is a diagram illustrating an algorithm based on a conventional edge detection device.

FIG. 2 is a block diagram showing the configuration of a line interpolation device using a conventional edge detection device.

FIG. 3 is a diagram illustrating a direction calculation algorithm according to the invention.

FIG. 4 is a block diagram showing the configuration of an edge direction detection device according to the present invention.

[Explanation of symbols]

22 Edge direction detection stage 24 Interpolator 26 Low pass filter 28 Multiplexer

Claims (9)

[Claims] 1. Apparatus for approximating the direction of the edges of a video image formed by a video signal consisting of sequential fields and sequential lines in each sequential field, the apparatus comprising: an edge detector comprising first means for generating a first signal that varies depending on the pixel value of the next pixel and the pixel value of an associated secondary pixel in another line; An edge detection device characterized in that the pixels are vertically adjacent pixels in a field subsequent to the current field. 2. The edge detection device according to claim 1, wherein the first means detects at least a pixel between a selected primary pixel and a selected secondary pixel constituting a first pixel pair. determining the absolute value of the difference in value and the absolute value of the difference in pixel values between the selected primary pixel and the selected secondary pixel constituting the second pixel pair; means for determining an angle with respect to the horizontal line determined by the first pixel pair that is different from an angle with respect to the horizontal line determined by the first pixel pair;
The difference signal between the pixel values constitutes the first signal, and further comprising evaluation means for receiving the first signal and generating an edge direction signal in response to evaluation of the first signal. Characteristic edge detection device. 3. Edge detection device according to claim 2, characterized in that the primary pixels form pixel groups each having at least three pixels. 4. The edge detection device according to claim 3, wherein the primary pixel constitutes a pixel group having first, second, and third pixels, respectively, and the corresponding secondary pixel is
forming a pixel group each having first, second, and third pixels that are mutually aligned in the vertical direction with the next pixel;
The signal has a first absolute value difference between the first pixel of the primary pixel and the third pixel of the secondary pixel, and a difference between the first pixel of the primary pixel and the second pixel of the secondary pixel. a second absolute value difference corresponding to the absolute value; a third absolute value difference between a vertically aligned pixel pair of a primary pixel and a secondary pixel; a third absolute value difference between a third pixel of the primary pixel and a third pixel of the secondary pixel a fourth absolute value difference between the two pixels, and a first representing the absolute value difference between the third pixel of the primary pixel and the first pixel of the secondary pixel.
An edge detection device characterized in that it has a difference signal of. 5. An edge detection device according to claim 1, wherein said first means includes means for generating a tangent signal. 6. Apparatus for generating an edge signal representative of the edge direction of a television picture constituted by pixels arranged in-line in a sequential field, the apparatus comprising: means for generating a first signal that partially varies; means for generating a second signal that varies at least partially in response to pixel values of horizontally aligned pixels; and means for generating a geometric function signal representative of the edge direction of the television image. 7. An edge detection device according to claim 6, wherein the vertically aligned pixels are in the same field. 8. An edge detection device according to claim 6, wherein the vertically aligned pixels are sequentially in a field. 9. An edge detection device according to claim 7, wherein a lookup is addressed by the geometrical function signal and displays an angle value representing an angle corresponding to the geometrical function signal. An edge detection device further comprising a table.