TW200939154A - An edge extraction and enhancement - Google Patents

Abstract

The present invention discloses an edge extraction and enhancement method, and particularly, an edge extraction and enhancement using Laplacian transform in horizontal, vertical, and diagonal direction as well as Laplacian cut with low noise may adjust the luminance of a pixel matrix yij, wherein said method can be fully implemented by hardware, embedded system or ASIC.

Description

200939154 IX. Description of the Invention: [Technical Field] The present invention relates to an edge extraction and strengthening method, and more particularly to a low noise which can be implemented using a Laplace operation and a Laplace cut using a hardware. Edge extraction and strengthening methods. [Prior Art] ® In digital imaging systems, a basic and important image processing technique is to extract the edges of the image to enhance the brightness of the edges of the image and enhance the blurred image to a clear image. In the conventional technique, it has the effect of strengthening the edge of the image regardless of the directionality. However, the above-mentioned operation method to capture the boundary of the object in the image has its shortcomings: although this method makes the edge of the object sharp, it also has the same effect on the noise in the image. Moreover, the edge data processed by the directional edge strengthening method is mostly fragmented and cannot provide complete object boundary data. U.S. Patent No. 6,463,175 discloses an edge extraction and enhancement method that uses idempotent processing and filter operations to reduce structure-guided image distortion, but the case needs to be implemented using a computer. Compared with the case, the hardware is implemented, the processing speed is slow, and more and more soft and hardware resources are needed to realize the edge extraction and strengthening method. Therefore, the inventor of the present invention has developed a simple and effective method of edge extraction and enhancement, which can be implemented by using a hard body, and the high-frequency part of the image 5 200939154 will be strengthened, and the low-frequency part will be attenuated. It can improve the current situation in the prior art that the processing speed is slow and more and more hardware and software resources are needed and the noise is amplified. [Summary of the Invention] The noise in the good knowledge technology is amplified. According to an exemplary example of the present invention, an edge extraction and enhancement method for adjusting the brightness value of a pixel yij is provided, which can be implemented by using a hardware. The present invention relates to an edge extraction and enhancement method. It uses Laplace Plass operation and Laplacian, which can be achieved by using hardware, horizontal, vertical,

The first edge value, the diagonal direction of the first image is Laplace, the operation has a horizontal value - the edge value is multiplied by a first gain and then the first second image.

Preferably, the first image is a matrix of 3×3, a variable of a negative direction; and an image is added to the image and then expressed as follows: Y′ = first image Yin + second edge value The matrix can represent 6 200939154 * is yij. Preferably, the Laplace operation is adjustable in horizontal, vertical, and diagonal directions. Preferably, the Laplace cut in the positive direction is adjustable. Preferably, the Laplace cut in the negative direction is adjustable. Preferably, the first gain is adjustable. Preferably, the first gain range is from 0 to 4. ® Preferably, the first image is 10 bits. Preferably, the hardware is an embedded system or a specific application integrated circuit. In order to enable the reviewing committee to have a better understanding and recognition of the structural purpose and efficacy of the present invention, the following examples are described in detail with reference to the illustrated examples. [Embodiment] FIG. 1 is a schematic diagram of an input image matrix of the present invention. The edge extraction system of the present invention extracts the edge of an image by pulling the input image in three directions of horizontal, vertical, and diagonal directions, and assigns the same or The different gains in turn control the degree of edge enhancement in the three directions. One pixel (such as y22) and its surrounding pixels in the image are exemplified by a Y matrix, such as: yn yi2 yi3 y2i m y23 ysi ys2 y33 Preferably, the Y matrix is a matrix of 3 by 3. 7 200939154 Y' is a matrix obtained by the arithmetic processing of the present invention.

LoY,, yn — _yj2 yis Yn ------ y23 yai ---- __X32 y33

Χν+ι^ car technology knows that the γ matrix can also be expressed as (2) the matrix of XJN + U, where Ν is a natural number. The W-like towel-pixel yij (such as Μ) and its surrounding pixels Take a 5 by 5 matrix γ matrix 1〇5 as an example.

Yn yi2 y.. y.. yzi y23 y.. v.. y3i _ys2 ysa y.. y.. y. · y.. y.. y.. y.. y.. y.. y.. Y.. y.. The edge extraction of the present invention extracts the image edge according to the horizontal, vertical and diagonal three sides of the image. Taking a 3 by 3 Y matrix as an example, the -y-edge value 201 can be expressed as: - edge

LoY - horizontal gain X (y22x2-y2i-y23) + vertical gain X (y22x2-yi2-y32) + diagonal gain X (y22x4-yn-yi3-y3i-y33); 200939154 • Those skilled in the art know, One pixel yi j (such as y22) and its surrounding pixels in the image are exemplified by a Y-by-N matrix, and the first edge value 201 can also be expressed as:

LoY = horizontal gain X (yN+i, tmx(2xN-l)-Eyi, jj=N+l) + vertical gain x (yN+i, N+ix(2xN-l)-Eyi, ji=N+l ) + diagonal gain x (yN+i, N+ix(2xN-l)-Eyi, jj=+/-i); Do the Laplacian operation of the input image to obtain the high-frequency term LoY, ® Information on the edge of the image. Here, the user can give horizontal, vertical, and diagonal gains by setting the horizontal gain, vertical gain, and diagonal gain to control the difference in edge enhancement between the three directions. 2 is a schematic diagram of edge enhancement according to the present invention. The present invention performs a Laplacian operation on the high-frequency term LoY of the obtained image, that is, the information value of the image edge, and obtains a numerical value to be adjusted for the edge of the individual image. LoY' is the second edge value 301, and the second edge value is added back to the same position as the original image to achieve the edge enhancement effect. The Laplace cut Ο equation can be represented as shown in Figure 2. The user can adjust the edge portion and the degree of enhancement to be achieved by setting: positive/negative high/low level (H / L). The positive and negative high/low parameters are designed to limit the edges of the image that really need to be enhanced to avoid noise enhancement. When L<LoY<H, LoY' =LoY-L; when -H<LoY<-L, LoY' = L〇Y+L; 9 200939154 when ~L<l〇Y<L, LoY,=〇; LoY, =hL; and when l〇Y$L, LoY, =-(hL); where H and L are the parameters of the middle and low level of the cleavage. Figure 2 is designed to 'based on the brightness of the original input image's luminance value'. We will give different edge enhancement gains EE_gain' for design convenience. It can be assumed that Y is a 1-bit input signal, ® edge enhancement The value of gain 301 (EE_gain) is between 0 and 4. The relationship between the enhanced image and the original image is: the second edge value is multiplied by a first gain and then added to the first image to output a second image. The equation for the relationship between the enhanced image and the original image can be expressed as follows: Y' = Y + LoY' X EE gain. It should be noted that the present invention has been described in the above specific embodiments, but is not limited thereto. Other variations can be made by those skilled in the art without departing from the scope of the invention. The edge extraction and strengthening method formed by the method of the invention, in particular, a method for edge extraction and strengthening using low noise of Laplace operation and Laplace cut, except that it can be realized by a hardware In addition, it can also be implemented as an embedded system or an application-specific integrated circuit (ASIC). Similarly, the field of application of the edge extraction and strengthening method of the present invention is not limited thereto. Persons having ordinary skill in the art 200939154 Other changes may be proposed by the priest, such as inputting Y as a 9-bit input signal without departing from the scope of the present invention. Referring again to Figure 4, Figure 4 is a flow chart of an example of an edge extraction and enhancement method in accordance with one of the present inventions. One pixel yij (such as y 2 2) and its surrounding pixels in the first image are represented by an N-by-N Y matrix 401, and the matrix 401 is subjected to a Laplace operation 402 to obtain a first edge value LoY. An edge value LoY is obtained by Laplacian 403 to obtain a second edge value LoY', and a second edge value LoY' is obtained by the equation 404 to obtain an output as a second image. Equation 404 can be expressed as follows: The edge value is multiplied by a first gain and then added to the first image to output a second image. Y, = Y + LoY, X EE_gain. "Examples of Implementation" In order to give your members a better understanding of the superiority of the present invention in practical use, the inventors will use eleven matrices as an example to perform operations according to the above, in which: EE gain = 3.5; and horizontal and vertical The diagonal gains in the three directions are: 4, 2, and 0; and in actual operation, Laplace cut can also be defined as the following equation: When EE_th < LOY < 127 + EE_th LoY ' = L0Y - EE —th; when -128-EE_th < LOY < -EE_th LoY' =L0Y+EE_th; when -EE_th <= LOY <= EE-th LoY' =0; 11 200939154 when LOY >= 127+EE_th LoY' = 127; and when LOY <= -128-EE-th LoY' = -128; where EE_th is the parameter of the high and low level of Laplacian. When the input matrix is represented by 501A (as shown in Figure 5A), one of the pixels yij (such as y22) in the image is operated as follows: LoY=4x(20-10-10)+2*(20-10-10)+ 0*(40-10-10-10-10)=0; LoY' =0; y22' =^22=10. ❹ When the input matrix is represented by 501B (as shown in Figure 5B), one of the pixels yu (such as y22) in the image is operated as follows: L〇Y=0; L〇Y’ =0; y22’ =y22=12. When the input matrix is represented by 501C (as shown in Figure 5C), one of the pixels yu (such as y22) in the image is operated as follows: LoY = 4; LoY' = 0; y22, = y22 = 12. When the input matrix is represented by 501D (as shown in Figure 5D), one of the pixels yu (e.g., y22) in the image is operated as follows: ❹ LoY = 8; LoY' = 4; y22 = 14; y22' = 28. When the input matrix is represented by 501E (as shown in Figure 5E), one of the pixels yu (such as y22) in the image is operated as follows: 1^(^=164(^:125 722=14: y22' =56 〇当输入The matrix is represented by 501F (Fig. 5F). One pixel yu (such as y22) in the image is operated as follows: LoY=8; LoY' = 4; y22=14; y22' = 28. When the input matrix is 501G Represented (Figure 5G) 12 200939154 One of the pixels yij (such as y22) in the image is operated as follows:

LoY=24; LoY’ = 20; y22=14; y"22 =84. When the input matrix is represented by 501H (as shown in Figure 5), one of the pixels yi j (such as y22) in the image is operated as follows:

LoY=16; LoY’ =12; y22=14; y〗 2,=56. When the input matrix is represented by 5011 (Figure 5 I)

之一 One pixel yu (such as yu) in the image is operated as follows:

LoY=32; LoY’ =28; y22=14; y22’ =112. When the input matrix is represented by 501J (as shown in Figure 5J), one of the pixels yi j (such as y22) in the image is operated as follows:

LoY=48; LoY’ =44; y22=14; y22’ =168. When the input matrix is represented by 501K (as shown in Figure 5K), one of the pixels yi j (such as yu) in the image is operated as follows: LoY = 8; LoY, = 4; y22 = 896; y22' = 903. The above is only the embodiment of the present invention, and the scope of the invention is not limited by the present invention. That is to say, the average change and modification of the scope of the invention according to the scope of the invention should still fall within the scope of the cover. Therefore, the present invention is actually a novelty, a goodness, and a letter to the industrial Xiangshu. (4) Please ask for the 妓 妓 可 审查 审查 审查 审查 ' 并 并 并 并 并 并 并 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 13 BRIEF DESCRIPTION OF THE DRAWINGS FIG. 3 is a schematic diagram showing the gain of an edge enhancement method according to the present invention; FIG. 4 is a schematic diagram showing an example of an edge extraction and strengthening method according to the present invention; and FIG. 5A to 5K. The input matrix diagram of the present invention. [Major component symbol description] 103-matrix 105-matrix 201-second edge value 301-edge enhancement gain 401-matrix G 402-lapar operation 403-lapras 404- equation 501A~K-matrix 14

Claims (1)

200939154 X. Patent application scope: 1. An edge extraction and enhancement method for adjusting the brightness value of a pixel yij, comprising: performing Laplacian operation on a first image in horizontal, vertical and diagonal directions. And outputting a first edge value LoY, the Laplacian operation has a horizontal gain, a vertical gain, and a pair of angular gains; performing a Laplacian on the first edge value and outputting a second edge value ❹ LoY', the Laplace cut has a positive and negative direction variable Η, L; and multiplies the second edge value by a first gain and then adds the first image to output a second image; The method is implemented in hardware. 2. The method of edge extraction and enhancement according to item 1 of the patent application, wherein the first edge value LoY = horizontal gain X (ypf+i, n+ix2xN-yN+i, N+i-Σ y I, jj= N+l) + vertical gain x (Υν+ι,ν+丨x2xN- yN+i,N+i-Zyi,ji=N+l) + ❹ diagonal gain x (ym,N+ix4xN- yN+i , N+i-Σ yi, jj=+/-i) ; N is a natural number. 3. The method of edge extraction and enhancement according to item 1 of the patent application, wherein the gain of the Laplace operation in the horizontal direction is adjustable. 4. The method of edge extraction and enhancement according to item 1 of the patent application, wherein the gain of the Laplace operation in the vertical direction is adjustable. 5. The method of edge extraction and enhancement according to item 1 of the patent application, wherein the gain of the Laplace operation in the diagonal direction is adjustable. 15 200939154 The edge extraction and strengthening method of item 1 of the range of interest, wherein the variables H and L of the razor cut in the positive direction are adjustable. It is the edge extraction and strengthening method of the first item of the patent scope of the patent. In the basin, the variable of the Lars, which is cut in the negative direction, H and _L are adjustable. ’... 9· ^Application for the edge extraction and strengthening method of Section 8 Poor The first gain range is 〇4. Bajia took the money from the first item of the coffee, and the brother-image was 10 yuan. 11: An edge extraction and enhancement method according to item 1 of the claim range, wherein the image is a 3×3 matrix, and the matrix can be expressed as y. 12. The method of edge extraction and enhancement according to item U of the patent application scope, wherein the edge extraction method has a Laplace operation (4) The horizontal gain x (y22x2, y23) + _ the vertical gain X (y22X2-y12-y32) + the diagonal gain X (y22x4-yii-yi3-y31-y33). 13. The method of edge extraction and enhancement according to claim 12, wherein the second edge value L0Y of the edge enhancement method is determined according to the following equation: when L<LoY< H, LoY, =LoY-L; when -H<LoY<-L ' LoY' =LoY+L; when -L<LoY<L,LoY,=〇; 200939154 when LoY^H, LoY' =HL; When LoYSL, LoY' = - (HL). 14. The edge extraction and strengthening method according to claim 12, wherein the second edge value L ο Y ' after Laplace cutting of the edge strengthening method is determined according to the following equation: when EE_th < LOY < 127 + EE - th LoY, = L0Y - EE_th; when -128-EE_th < L0Y < -EE_th LoY' = L0Y + EE_th; when -EE_th <=LOY <= EE_th LoY, =〇; LOY >= 127+EE_th LoY' = 127; and when LOY <= -128-EE_th LoY, =-128 . 15. For the edge extraction and strengthening method of claims 1 to 14, where the hardware is a diffused system. 16. For the edge extraction and strengthening method of claims 1 to 14, wherein the hardware is a specific application integrated circuit. 17