CN1553708A - Method for detecting dynamic image pixel with adjustable critical value - Google Patents

Abstract

The present invention is a dynamic image pixel detection method with adjustable critical value. It defines a variable critical value through the information of dynamic pixels in the image, and uses it to judge whether the missing pixels in the image are located in a static region or a non-static region. If the dynamics of the missing pixel adjacent to the reference field are large, a small critical value is used; if the dynamics of the missing pixel adjacent to the reference field are small, a larger critical value is used to eliminate the erroneous judgment caused by noise, and the purpose of optimally reconstructing the missing pixels is achieved by using inter-field interpolation and intra-field interpolation respectively.

Description

Dynamic Image Pixel Detection Method with Adjustable Threshold Value

technical field

The present invention is a dynamic image pixel detection method, which judges whether the missing pixel in the image is located in a static area or a non-static area through a variable critical value, so as to eliminate the wrong judgment caused by noise, and achieve reconstruction of missing pixels by interpolation The purpose of the pixel.

Background technique

The text of the manual.

In the prior art, a dynamic adjustment algorithm (motion-adaptive algorithm) judges whether a missing pixel in an image is located in a static area or a non-static area based on the difference between surrounding adjacent pixels. Afterwards, if it is judged that the missing pixel is located in a static area, the missing pixel will be reconstructed by an inter-field interpolation method using the pixel information of adjacent fields (neighboring fields); if it is judged that the missing pixel is located in a non-static area, The missing pixel is reconstructed by intra-field interpolation of information from neighboring original scan lines in the same field.

Please refer to FIG. 1A for a schematic diagram of the front and back field interpolation method in the prior art. Wherein the missing pixel (missing pixel) 10 occurs at the X shown in the figure, and the front time pixel 11 and the back time pixel 13 of the missing pixel 10 shown in ○ are originally existing pixels, and the coordinates of this pixel at each time point is (x, y), the front time pixel 11, the missing pixel 10 and the back time pixel 13 are displays at three different times. The image information of the pixel 13 is averaged to obtain the reconstruction value of the missing pixel 10, namely:

$\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

Wherein, F(x, y, n−1) is the pixel function formula of the previous pixel 11 , and F(x, y, n+1) is the pixel function formula of the next time pixel 13 .

Please refer to FIG. 1B for a schematic diagram of the surrounding field interpolation method in the prior art. Wherein the missing pixel 10 is located at the × shown in the figure, and the first position pixel 14 and the second position pixel 16 shown in ○ are pixels that originally existed, and its first position pixel 14, missing pixel 10 and second position pixel 16 The coordinates of are respectively (x, y-1), (x, y) and (x, y+1), which are the display of three upper and lower pixels. If the missing pixel 10 is to be reconstructed, the first The image information of the position pixel 14 and the second position pixel 16 is averaged to obtain the reconstruction value of the missing pixel 10, namely:

$\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\frac{<span\; class="notranslate">\; [</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ]</span>}{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

Wherein, F(x, y−1, n) is the pixel function formula of the pixel 14 at the first position, and F(x, y+1, n) is the pixel function formula of the pixel 16 at the second position.

To determine whether the missing pixel 10 is located in a static or non-static area, it is judged by using the information difference of adjacent pixels around the missing pixel 10. If the difference value is less than a threshold (threshold), it means that the pixels around the missing pixel 10 are static, and it is also judged The missing pixel 10 is static; if the difference value is greater than or equal to the critical value, the surrounding pixels are non-static, that is, it is determined that the missing pixel 10 is located in a dynamic area.

Please refer to FIG. 2 for a schematic diagram of a method for calculating difference values in the prior art. This figure shows the calculation of the above-mentioned difference value, using the front time field 21 of the missing pixel 20 shown as × in the figure and the surrounding non-missing pixels shown as ○ in the back time field 23 to calculate the absolute value sum of the difference ( sum of absolute difference, SAD), that is, the interpolation method of the front and rear time fields in Figure 1A above and the surrounding interpolation method of the upper and lower positions in Figure 1B are used in conjunction, and the calculation formula is as follows:

$\mathrm{<span\; class="notranslate">\; Diff</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; \&gamma;\&Gamma;</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

$\mathrm{<span\; class="notranslate">\; \&Gamma;</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; \{</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; ,</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \}</span>$

Among them, Diff(x, y, n) is the difference value function formula, ∑|f(i, j, n-1)-f(i, j, n+1)| is calculated for the surrounding pixels of the time field before and after the missing pixel 20 The absolute value sum of the difference, the functional formula f (i, j, n-1) represents the previous time field 21 shown in Fig. 2, and the functional formula is each pixel function in F (n-1), including each pixel Information, f(i, j, n+1) represents the post-time field 23, and the functional formula is the function of each pixel in F(n+1), while (i, j) represents the position of the front and back time field pixels, The positions include (x, y-2), (x, y), (x, y+2), (x-1, y) and (x+1, y).

FIG. 3 is a flowchart of a dynamic pixel adjustment operation in the prior art. As shown in the figure, F(n) is a time field function formula. When the dynamic pixel adjustment operation starts, the time field function formula F(n) is calculated by deinterlacing (step 301), and a deinterlacing processor ( deinterlacingprocessor) input time field F (n), front time field F (n-1) and a function formula is the back time field (step 303) of F (n+1), after that, scan out the time field F in the image ( n), and perform interpolation (step 305) in a raster order (raster order), that is, sequential scanning from top-left to bottom-right (top-left to bottom-right) of the image, and then The calculation formula described in 2 calculates the difference value of the pixels surrounding the missing pixel (step 307), and compares it with a critical value to determine whether the missing pixel is located in a static or non-static area (step 309), if the difference value is smaller than the critical value, Then rebuild with the front and back field interpolation method described in FIG. 1A (step 311), if the difference value is greater than or equal to the critical value, then rebuild with the surrounding field interpolation method described in FIG. 1B (step 313), and then output the reconstructed solution Interleave pixels (step 315), then judge whether to end the de-interleaving operation of this field function F(n) (step 317), if not, continue to scan missing pixels, and perform interpolation 305 in a raster sequence until time The missing pixels in the field F(n) are completely reconstructed; if yes, the de-interleaving operation of the field function F(n) is ended (step 319 ).

The above is the operation process of dynamic pixel adjustment in the prior art. This method is simple and easy to implement, but only using a fixed critical value to judge whether the missing pixel is located in a static or non-static area may cause regional changes in the image to be regarded as regional changes. Wrong judgment, such as the noise caused by image format conversion, such noise may cause the difference value of adjacent pixels to be greater than the critical value, and the missing pixels originally located in the static area should use the front and back field interpolation method, and use the surrounding The intra-field interpolation method wrongly reconstructs the missing pixels. If there are only one or two missing pixels due to wrong judgment, the reconstruction will not have much impact. If there are many pixels to be reconstructed, it will cause a considerable degree of image flickering.

In order to solve the above-mentioned erroneous judgment of regional changes in the image caused by the use of a fixed critical value, resulting in erroneous reconstruction of missing pixels in the image, the present invention proposes a dynamic image pixel detection method with an adjustable critical value to achieve correct Static and non-static area judgment.

Contents of the invention

The invention is a dynamic image pixel detection method with adjustable critical value. A variable critical value is defined by the information of dynamic pixels in the image, and the difference between the adjacent pixels of the missing pixel and the surrounding pixels is compared with this critical value to determine whether the missing pixel in the image is located in a static area or a non-static area. It is also possible to eliminate erroneous judgments caused by noise by changing the critical value, and then achieve the best reconstruction of missing pixels by the front and back field interpolation method and the surrounding field interpolation method.

In order to achieve the above object, the dynamic image pixel detection method with adjustable critical value of the present invention is to calculate the time field where a missing pixel is located by deinterlacing, and input the time field and a plurality of reference fields in a deinterlacing processor, and then Calculate the change amount of the reference field, and then determine a critical value according to the change degree of the reference field, scan out the missing pixel and calculate the difference value of the surrounding pixels around the missing pixel, and then compare it with the previously determined critical value to judge the previous The missing pixels are reconstructed by back field interpolation or by surrounding field interpolation. Finally, the reconstructed de-interlaced pixels are output, and it is judged whether to end the de-interlaced operation of the time field or to repeat steps to achieve the purpose of reconstructing missing pixels.

Description of drawings

Relevant detailed content and technology of the present invention, now refer to accompanying drawing and explain as follows:

FIG. 1A is a schematic diagram of front and back field interpolation methods in the prior art;

FIG. 1B is a schematic diagram of the surrounding field interpolation method in the prior art;

Fig. 2 is a schematic diagram of a method for calculating difference values in the prior art;

Fig. 3 is a flow chart of dynamic pixel adjustment operation in the prior art;

FIG. 4 is a flowchart of a dynamic pixel adjustment operation using a variable threshold value according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of defining a variable threshold value according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of a reference field of a missing pixel in an embodiment of the present invention;

FIG. 7 is a schematic diagram of a border-guided surrounding image frame interpolation method according to an embodiment of the present invention;

FIG. 8 is a flowchart of a dynamic image pixel adjustment operation according to an embodiment of the present invention.

Symbol Description

10....................missing pixels;

11....................Pixels of previous time;

13....................back time pixels;

14..............................first position pixel;

16..............Second position pixel;

20........... missing pixels;

21....................Previous time field;

23 .................... post-time field;

61.................................First time field;

63....................Second time field;

65.....................the third time field;

67....................................Fourth time field;

70..............missing pixels;

71.....................first neighboring pixel;

72....................Second adjacent pixel;

73.....................the third adjacent pixel;

74.....................the fourth adjacent pixel;

75....................Fifth Neighboring Pixel;

76...................Sixth adjacent pixel;

D1..................... the first difference value;

D2................................Second difference value;

Th1.....................First critical value;

Th2....................................Second critical value;

U1.....................the first difference;

U2................................Second difference;

U3.....................the third difference;

Diff.................................difference value;

FieldDiff.............Field difference value.

Detailed ways

The present invention provides a method for detecting whether an image pixel is static or non-static with an adjustable threshold value, which will be described below through the flow chart of the embodiment.

FIG. 4 is a flowchart of a dynamic pixel adjustment operation using a variable threshold value according to an embodiment of the present invention. A variable critical value is defined by the information of dynamic pixels in the image, and the difference between the surrounding pixels of the missing pixel (missing pixel) is compared with this critical value to determine whether the missing pixel in the image is located in a static area or a non-static area The judgment of the area, the missing pixels are the pixels discarded when the de-interlaced image is converted from the image frame to the time field (frame-to-field conversion). As shown in the figure, F(n) is a time field function formula. When the dynamic pixel adjustment operation starts, the time field function formula F(n) is deinterleaved and calculated (step 401), and a deinterleave processor inputs the time field F (n) and adjacent surrounding reference fields (step 403), after that, calculate the variation of adjacent surrounding reference fields (step 405), and define an appropriate critical value with dynamic adjustment (step 407) (the method is described in the embodiment of the present invention Fig. 5 explains in detail), then scan out the missing pixels in the time field F(n) in the image, and perform interpolation (step 409) in a raster sequence, that is, from the upper left to the lower right of the image (top- left to bottom-right) sequential scanning.

In order to calculate whether the area around the missing pixel is static or non-static, the formula (3) described in Figure 2 $(\mathrm{Diff}(x,\mathrm{the\; y},\mathrm{no})=\underset{(i,j)\&Element;\mathrm{\&Gamma;}}{\mathrm{\&Sigma;}}|f(i,j,\mathrm{no}-1)-f(i,j,\mathrm{no}+1)\left|\right)$ To calculate the difference value Diff (step 411) of surrounding pixels around the missing pixel, if the difference value is large, then it is judged that its plurality of reference fields are high-fluctuation images; if the difference value is small, then it is judged that the plurality of reference fields are low Change the image, and judge whether the missing pixel is located in a static or non-static area by comparing it with a critical value, that is, judge whether the difference value Diff is less than the critical value (step 413), if it means that the difference value Diff is less than the critical value, then use The inter-field interpolation method (inter-field interpolation) described in FIG. 1A reconstructs missing pixels (step 415); if not, it means that the difference value Diff is greater than or equal to the critical value, and the surrounding field interpolation method described in FIG. 1B ( intra-field interpolation) reconstruction (step 417), then output the reconstructed de-interlaced pixels (step 419), then judge whether to end the de-interlacing operation of the field function F(n) (step 421), if not, continue scanning Missing pixels, and perform interpolation in a raster sequence (step 409), until the missing pixels in the time field F(n) are completely reconstructed; if so, it means that the de-interlacing operation of this field function formula F(n) is ended (step 423 ).

Please refer to FIG. 5 , which is a schematic diagram of defining a variable threshold value according to an embodiment of the present invention, which is the steps of defining a variable threshold value (step 405 to step 407 ) shown in the flow chart of FIG. 4 . First calculate the amount of change (step 405) of the missing pixels adjacent to the surrounding reference field pixels of the scan, and judge whether the dynamics of the reference field is large, which is a high-change image (step 501), if so, because of the masking effect (masking effect) ), the noise in the image has little influence relative to the high-variation image, so it is selected to use a small critical value (step 503); if not, the reference field dynamics is small, and it is a low-variation image. Due to the facilitation effect (facilitation effeet), the influence of the noise in the image appears to be large in the low-variation image, so a large critical value should be selected (step 505). The critical value selected above is the flow of the dynamic pixel adjustment operation.

Please refer to FIG. 6 for a schematic diagram of the reference field of the missing pixel in the embodiment of the present invention. The difference between the adjacent reference fields of the missing pixel is used to calculate the amount of change. Wherein, if the missing pixel (missing pixel) is located in the upper field, that is, the third time field 65 shown in the figure Inside, that is, the current time field, marked with X, and its reference field is the adjacent time field. As shown in the figure, the plurality of reference fields and the current time field, the first time field 61, the function formula is F(n-2), the second time field 63, the function formula is F(n-1), and the missing pixels are in In the third time field 65, the function formula is F(n), and in the fourth time field 67, the function formula is F(n+1). In this embodiment, the difference between the four time fields is judged as a high-change graph If the image is still a low-variation image, its field difference value FieldDiff uses the sum of the absolute values of the difference values (SAD) between its front and back fields, and the calculation formula is as follows:

$\mathrm{<span\; class="notranslate">\; FieldDiff</span>}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 0</span>}}{\overset{\mathrm{<span\; class="notranslate">\; N</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span>$

Among them, F(n) and F(n-2) are the upper field (top field), F(n-1) and F(n+1) are the lower field (bottom field), and M and N are the images in the image respectively. The height and width of the frame, the symbols of each original pixel and missing pixel shown in the figure are ○ and × respectively, and the position coordinates are expressed as (x, y) and so on. The threshold value is defined by the field difference value FieldDiff, and the threshold value size is determined by comparing another dynamic threshold value with this field difference value FieldDiff. When the field difference value FieldDiff is greater than or equal to a preset dynamic threshold value, it is judged The images of these plural time fields are high-variation images, and it is determined as a small critical value; the difference value FieldDiff on the spot is smaller than the dynamic critical value, and then it is judged that the images of these plural time fields are low-variation images, and it is determined as a low-variation image. large critical value.

Please continue to refer to FIG. 6 for a schematic diagram of difference value calculation in an embodiment of the present invention. In order to improve the accuracy of judging whether the image is static, the present invention uses four time fields in the upper field and the lower field in the image frame to calculate the difference value, which is the first time field 61 respectively, and the function formula is F(n- 2), the second time field 63, the function formula is F(n-1), the third time field 65 where the missing pixel is located, the function formula is F(n), the fourth time field 67, the function formula is F(n+ 1). The field to be deinterlaced is the third time field 65 with missing pixels. If the current time field is a top-field, the reference field has a plurality of front time fields and a back time field. As the complex number of reference fields required for calculating the difference value, the reference fields based on the third time field 65 include: the first time field 61 is the time field of the previous state, and the second time field 63 is the time field of the previous state. The fourth time field 67 is the time field of the subsequent state. And calculate the first difference value D1 by each pixel position of the first time field 61 and the third time field 65; Calculate the second difference value D2 by each pixel position of the second time field 63 and the fourth time field 67, its operation The formulas are as follows:

$\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; \&Gamma;</span>}\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

$\mathrm{<span\; class="notranslate">\; D.</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; \&Gamma;</span>}\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}<span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; no</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; |</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

Among them, Γ1 and Г2 are pixel positions, respectively as follows:

Г1=[(x-1, y-1), (x, y-1), (x+1, y-1), (x-1, y+1), (x, y+1), ( x+1, y+1)]

Г2=[(x, y-2), (x, y), (x, y+2), (x-1, y), (x+1, y)]

In addition, if the current time field where the missing pixel is located is the bottom-field, the remaining reference fields for calculating the difference include: a time field of the previous state and a plurality of time fields of the subsequent state.

FIG. 7 is a schematic diagram of a frame interpolation method for boundary-guided surrounding images according to an embodiment of the present invention. The boundary-guided surrounding interpolation method is to select the pixel with the smallest functional difference between the pixels at the diagonal positions around the missing pixel, and then the average value of the pixels at the diagonal positions is the reconstruction value of the missing pixel. As shown in the figure, the symbol X represents a missing pixel 70, and its adjacent pixels include a first adjacent pixel 71, a second adjacent pixel 72, a third adjacent pixel 73, a fourth adjacent pixel 74, a fifth adjacent pixel 75 and a sixth adjacent pixel. Adjacent pixels 76 are used to determine whether the missing pixel 70 is located in a static or non-static area by using the difference between corresponding adjacent pixels.

Among them, the first difference U1 represents the difference between the first adjacent pixel 71 and the sixth adjacent pixel 76; the second difference U2 represents the difference between the second adjacent pixel 72 and the fifth adjacent pixel 75; the third difference U3 represents the third The difference between the adjacent pixel 73 and the fourth adjacent pixel 74 is found, and the smallest difference is found to be used as the judgment of the static pixel and the non-static pixel. If the minimum value is U1, as shown in the figure, the missing pixel 70 is the average of the first adjacent pixel 71 and the sixth adjacent pixel 76; if the minimum value is U2, the missing pixel 70 is the average of the second adjacent pixel 72 and the fifth adjacent pixel. The average of the pixels 75; if the minimum value is U3, the missing pixel 70 is the average of the third adjacent pixel 73 and the fourth adjacent pixel 74.

After that, the reconstruction value of the missing pixel 70 is further obtained by an intermediate value filtering method. This intermediate value filtering method is to find the intermediate value of the missing pixel and the upper and lower relative pixels as the missing pixel reconstruction value, and to reduce the reconstruction time of the missing pixel. The error method, in this embodiment, is to find the mean of value of the second adjacent pixel 72 and the fifth adjacent pixel 75, and use the mean of value as the new reconstruction value of the missing pixel 70. The above is the implementation content of edge-oriented intra-frame interpolation and median filtering.

Please refer to Fig. 8 for the dynamic image pixel adjustment operation flow chart of the embodiment of the present invention, which is the first difference value D1 and the second difference value D2 calculated from the difference between the fields shown in Fig. 6, and set two To determine whether the missing pixel is static or non-static, the corresponding and variable critical value can be used to determine the critical value of the change through a variety of built-in change conditions in a look-up table. The dynamic image pixel adjustment operation flowchart is as shown in Figure 8, F (n) is a time field function formula, when this dynamic pixel adjustment operation flow process starts, this time field function formula F (n) is deinterleaved calculation (step 801 ), a deinterleaving processor inputs each time field functional formula (step 803), is respectively the first time field 61, and the functional formula is F(n-2), the second time field 63, and the functional formula is F(n-1 ), the third time field 65, the fourth time field 67 whose function formula is F(n) and the function formula is F(n-1). Afterwards, use the difference between time fields shown in Figure 6 to calculate the field difference value between each field (step 805), and compare whether it is larger than a dynamic critical value set separately (step 807), in order to judge the missing The image where the pixel is located is a high-variation image or a low-variation image, if so, then define a small first critical value and a second critical value (step 809); if not, then define a first critical value and a second critical value is large (step 811).

Afterwards, continue to scan out the missing pixels in the image time field F(n) and perform interpolation in a raster sequence (step 813), calculate with the difference value described in Fig. 2, and pass each field shown in Fig. 6 Difference calculation of the first difference value D1 and the second difference value D2 (step 815), respectively, the first time field 61 and the third time field 65 calculate the first difference value D1, and the second time field 63 and the fourth time field 67 Calculate the second difference value D2, then use the previously defined corresponding first critical value Th1 and second critical value Th2, and compare the two critical values with the two difference values to determine whether the missing pixel is located in static or non-static Static area (step 817), if the first difference value D1 is less than the first critical value Th1 and the second difference value D2 is smaller than the second critical value Th2 at the same time, it is judged that the missing pixel is located in the static area. Intra-field interpolation reconstructs missing pixels (step 819). When the first difference value D1 is greater than or equal to the first critical value Th1 or the second difference value D2 is greater than or equal to the second critical value Th2, it is determined that the missing pixel is located in the dynamic region, and the missing pixel is reconstructed by the surrounding field interpolation method in the prior art ( Step 821), the surrounding field interpolation method further includes two steps, first, the boundary-guided surrounding image frame interpolation method described in FIG. 7 is used to reconstruct the missing pixel (step 823), which will not be described in detail here. Then a median filtering method is used to obtain the reconstruction of the missing pixels (step 825). Then output the reconstructed de-interlaced pixels (step 827), then judge whether to end the de-interlaced operation of the field function F(n) (step 829), if not, continue to scan the missing pixels and perform interpolation in a raster sequence (Step 813 ), until the missing pixels in the temporal field F(n) are completely reconstructed; if yes, it means that the de-interleaving operation of the field function F(n) is ended (Step 831 ).

The above is the detailed description of the embodiment of the dynamic image pixel detection method with adjustable critical value of the present invention. The present invention uses the information of the image around the missing pixel to define the variable critical value, so as to improve the detection of the static and non-static regions where the missing pixels are located. Static judgment to correctly reconstruct missing pixels. In addition, a boundary-guided surrounding image frame interpolation method and median filtering method are used for more stringent reconstruction.

To sum up, the dynamic image pixel detection method with adjustable critical value of the present invention is extremely advanced in terms of purpose and efficiency, has great industrial utilization value, and is an unprecedented new invention in the market , in full compliance with the requirements of invention patents.

The above descriptions are only preferred embodiments of the present invention, and should not be used to limit the implementation scope of the present invention. All equivalent changes and modifications made according to the patent scope of the present invention should still belong to the scope covered by the patent of the present invention.

Claims (10)

1. A dynamic image pixel detection method with adjustable critical value, used in the method of reconstructing missing pixels, the missing pixels are located in a time field, and the reconstruction steps include: input the time field and a plurality of reference fields; calculating the variation of the plurality of reference fields; Defining a critical value is to determine the critical value according to the variation of the plurality of reference fields; scanning the missing pixels and performing interpolation in a raster sequence; Calculate the difference value of the surrounding pixels surrounding the missing pixel; Comparing the difference value with the critical value; If the difference value is less than the critical value, the missing pixel is reconstructed using front and back field interpolation; If the difference value is greater than or equal to the critical value, the missing pixel is reconstructed using surrounding field interpolation; outputting the reconstructed de-interlaced pixels; and It is judged whether to end the deinterleaving operation of the time field. 2. The dynamic image pixel detection method with adjustable threshold value as claimed in claim 1, wherein the missing pixel is a de-interlaced image when the image frame is converted to a time field (frame-to-field conversion) The discarded pixels. 3. The dynamic image pixel detection method with adjustable critical value as claimed in claim 1, wherein the change amount of the missing pixel adjacent to the surrounding reference field is judged, if it is a high-change image, it is defined as a small critical value, if A low-variation image is defined as a large threshold value. 4. The dynamic image pixel detection method with adjustable critical value as claimed in claim 1, wherein if the difference value is less than the critical value, the missing pixel is reconstructed by front and back field interpolation; if the difference value is greater than or equal to the critical value, the missing pixel is reconstructed by surrounding field interpolation. 5. The dynamic image pixel detection method with adjustable threshold value as claimed in claim 1, wherein the surrounding interpolation method averages the image information of a pixel at a first position and a pixel at a second position to obtain the missing Method for reconstructing values from pixels. 6. A dynamic image pixel detection method with adjustable critical value is a method for reconstructing missing pixels in the current time field, and the reconstruction steps include: inputting the current time field and a plurality of reference fields; calculating field difference values between the current time field and the plurality of reference fields; comparing whether the field difference value is greater than a dynamic threshold; If so, define the first critical value and the second critical value to be small; If not, defining the first critical value and the second critical value to be large; Scanning out the missing pixels and performing interpolation in a raster sequence; calculating a first disparity value and a second disparity value of pixels surrounding the missing pixel; comparing the first difference value, the second difference value with the first critical value, the second critical value; If the first difference value is less than the first critical value, and the second difference value is less than the second critical value, then use front and back field interpolation; If the first difference value is greater than or equal to the first critical value, or the second difference value is greater than or equal to the second critical value, then use the surrounding field interpolation method; outputting the reconstructed de-interlaced pixels; and It is judged whether to end the deinterleaving operation of the time field. 7. The dynamic image pixel detection method with adjustable threshold value as claimed in claim 6, wherein the first difference value and the second difference value are a plurality of surrounding pixels of the missing pixel in the front time field and the back time field The sum of the absolute values of the differences. 8. The dynamic image pixel detection method with adjustable threshold value as claimed in claim 6, wherein the field difference value is the sum of the absolute values of the difference values between the preceding and following fields. 9. The dynamic image pixel detection method with adjustable threshold value as claimed in claim 6, wherein the change amount of the missing pixel adjacent to the surrounding reference field is judged, and if it is a high-fluctuation image, it is defined as the small first critical value and the second critical value, if it is a low-motion image, then define the first critical value and the second critical value as large. 10. The dynamic image pixel detection method with adjustable threshold value as claimed in claim 6, wherein the surrounding field interpolation method comprises a boundary-guided surrounding field interpolation method and an intermediate value filtering method.

Images