TW200539052A - An unarmed type 3D image rebuilding and distortion compensating method - Google Patents

Abstract

An unarmed type 3D image rebuilding and distortion compensating method is provided. Gather several 3D image pixels from unarmed-scan 2D image data. Then, use directly stack-based to continuously display the 3D image pixels and to re-build the 2D image data into 3D image data. Besides, use scale adjustment to achieve the effect of distortion compensation. Therefore, the time wasting and image distortion issues of re-building 2D image data into 3D image data are improved. Thus, users can judge the image features correctly.

Description

200539052 V. Description of the invention (1) [Technical field to which the invention belongs] The present invention relates to a variety of free-hand image reconstruction blood, in particular, a method using direct stacking method to perform three-way; J compensation method, and a method of adjusting scale for distortion compensation . Zhunru image reconstruction and the use of the previous technology] The technique 'reconstructs the reconstruction and rendering tasks with the assistance of the dry and mechanical three-dimensional positioning system through the reconstruction of the known three-dimensional ultrasound system of magnetic phase-dimensional image data, because Requires additional positioning aids: It seems that 丨 no shell, 丨 κ is more expensive, and the software processing speed is faster than the overall equipment display effect. ',' ,, and more, can not be satisfied, and for the operation of the three = ultrasonic system (eg, ultrasonic instruments), the value system is not required, and the hands are held with bare hands. Scanning the object to be measured with a wave probe, you can get 3D image data with Micro Motion / + on the display screen, which will make the operation process more convenient, but Tian Er ^ 1 ¾ freehand scan · operation speed during sighting May be inconsistent, so the three-axis scale may not change a, so it is extremely common that freehand scanning causes image distortion. The modification of _ is based on the above situation. Although there are speckle Tracki R egistrati ◦ image reconstruction techniques such as η that do not require a three-dimensional positioning system. § One, the algorithm of the above technology requires a longer processing time, or / but the device speeds up processing. Speed virtually increases the cost of the operating system with the aid of hard and inspection equipment. The complexity of personnel and how to use the free-hand scanning of the two-dimensional image as three-dimensional image data, and to reduce the distortion to a minimum is a common goal of researchers and researchers. nine

200539052 V. Description of the invention [Invention of the three-dimensional image of the original image reconstruction software. So -dimensional image and threshold value determine several values. 'Setting the aforementioned three-dimensional image formation technique, which is to adjust the image recognition is not as simple as: Take the large pixel value to produce the whole value (Sf each axis is newly based on each axis

Product of X dt Product (2) Content] The purpose of Ming is to solve the above problems, and propose a method of free-standing tri-construction and distortion compensation. It uses the direct stacking method to shorten the processing time, so as to reduce the time for restacking and drawing the 2D image data into breeding materials. In order to achieve the above object, the present invention includes the following steps: continuously extracting two-dimensional pixel values (pixel) in the data, by using the two-dimensional pixel values and a specific calculation method (such as the nearest neighbor interpolation algorithm), the closest of the three-dimensional points Adjacent points; capturing two-dimensional pixels of the nearest neighbor points as the voxel value of the three-dimensional point; and sequentially stacking the pixel values into a three-dimensional image data. Another purpose of Ming is to provide three-dimensional image distortion compensation technology by setting the ratio of the three axes to be consistent, and then according to the scale of each axis to compensate, to produce a more realistic image, and reduce the regret of shadow errors and effectively improve The accuracy of the image display. For the purpose of the above-mentioned distortion compensation, the present invention further includes the following steps (X dt, y dt, z d1) in the two-dimensional image of the number of pixels (X dt, y dt, z d1), the maximum number of pixels (md) and M The ratio of the values produces the corresponding scale adjustment value (sa); and then adjusts according to the scale) and the number of pixels on each axis (X dt, y (H, Z dt)) to produce a number of pixels (X d2, y d2, z d2) Perform the adjustment procedure; then multiply the number of new pixels (X d2, y d2, Z dZ) and the number of pixels in each axis (y dt, Z dt) by the ratio of the original scale (X sl, y sl, z sl) New scales (x s3, y s3, z s3): Finally, according to the new scale

Page 7 200539052 • ............— __ V. Description of the invention (3) | Ruler (ys3, zs3) and the original scale | Point interpolation algorithm to generate a new three-dimensional image II zsl) Use the latest For the purpose of the present invention: lean material. The solution is explained as follows: $ and its functions are further implemented.] I compensation; i "f :: means: second-hand 3D image reconstruction and distortion 丨 the way of 3D image *, when the user through scanning Head: Two: 7-dimensional Λ image Λ built as a & > ι 缶 + f, heart w _Mada Yoshi (m oda 1 1 tv) order eight-dimensional imaging system 1 〇 (eg, walking咅 y 川川 合 k «^ 20 ^ # | ^ ^ ^ ^ ^ ^ ^ ^ ^ # | Τ / Znλ A (\ JU 心心 1 豕 贝 料, or hunting by a computer; ^ Image management system 5 〇 Extract digital digital two-dimensional image data: order 1 Weidan m2 modeling group 3 ° image stacking unit in order to take two-dimensional pixel values (Plxel) in order to set a number of standings ^ Who crystal Zhi 盍: 脰The three-dimensional pixel value (voxel) of the point becomes a three-dimensional shadow silly ancestor through direct crystal stacking method. 隹 隹 should pass the image display unit two r can f, the quality of the shirt image data can be transparent through the image data can also be through 2 2 The order platform 40 is confirmed, and the other three | management system 5. Medium. The source of the two-dimensional image data stored in the digital image tube through the "executable platform 4Q can also be obtained through 40 through ㈣60. The above The volume of points = the number of two-dimensional pixels in the row platform xel), 1 using the one-dimensional image data to determine the f-nearest neighbor interpolation algorithm) in the fixed gate ^ value and several stereo points, The captured value of the distance between the nearest neighbors of the stereo points is set as the stereo pixel value of the stereo points. The n-threshold value is a two-dimensional pixel of $ | ′, knowing and taking three-dimensional. Page 8 200539052 V. Description of the invention (4) The number of points. Please refer to the "Figure 2" schematic diagram of interpolation calculation of two-dimensional image data to explain the method of selecting the nearest neighbors of three-dimensional points. P0 to P8 in the "original scale" are the original 9 pixels (including P0) in the two-dimensional image data. When the scale is reduced, you can scan P 0 to P 1 6 and the number of pixels will increase to 17 (Including P0). And when the scale is enlarged, you can scan P0 to P4, and the number of pixels will be reduced to 5 points (including P0). Next, the distortion compensation method is explained. When the user confirms that the three-dimensional image data needs to perform distortion compensation through the computer-executable platform 40, the three-dimensional image data can be successively transmitted to the image distortion compensation unit 3 in the three-dimensional image reconstruction module 30. 3. This image distortion compensation unit 3 3 will perform a series of calculations on the distorted parts in the 3D image data (such as the distortion compensation calculation and the nearest neighbor interpolation algorithm), and make appropriate adjustments to the scale of each axis to obtain high quality 3D image data, the quality of this 3D image data can be confirmed by the display device, and the adjusted 3D image data can be stored in the digital image management system 50 through the computer-executable platform 40, or through the network 6 0 is transmitted to other units. Please refer to the "3rd figure", "4th figure" and "5th figure" reconstruction method of freehand 3D image, in which "3rd figure" is the operation flowchart of the 3D image reconstruction method; "4th figure" This is a flowchart of capturing images in the reconstruction method, explaining the method of capturing 2D images in the workflow of the 3D image reconstruction method; and "Figure 5" is the direct stacking used in the freehand 3D image reconstruction method The operation flow chart of the method describes the operation flow of stacking two-dimensional image number (pi xe 1) into three-dimensional image data. The following description

Page 9

V. Description of the invention (5) The operation flow: First, set the number of two-dimensional pixels of the two axes X and Y in the two-dimensional image data (step 2 0); when the source of the two-dimensional image is from a two-dimensional imaging system ' Detect the probe depth of the probe of the 1D imaging system 10 (step 20 1), the scale of the X and Y axes (step 202), and take a few seconds of the image after the probe's moving distance (step 2 0 3), then press the image capture button (step 204), and smoothly move the probe of the 2D image capture module 20 in the 2D imaging system, so that it is within the set time and set distance. The two-dimensional image gas values in the two-dimensional image data are continuously acquired (step 205), and the manipulated two-dimensional pixel values are stored in the number 50 through the computer-executable platform 40 (step 206). y 1 豕 & buried system, and then, according to the two-dimensional pixel meal, ". dot (step 210); from the image data interface: X ::, the gate value determines the neighboring 5 0) to capture two-dimensional images In the data, σ, the image management system 311), this two-dimensional pixel value will be based on the pixel value (Puel) (in the two-dimensional image memory area (step 3 丨 sub 〇 preset image stick value in the image stacking unit 31). With the specific operator's image stack unit 3 1 will determine the nearest neighbor of the stereo point according to the ★ nearest neighbor interpolation algorithm) Then, according to the nearest neighbor ': Step 313). Step 22 0); Pixel value setting Stereo pixel value Stereo pixel value (V0Xel) (The two-dimensional pixel value of the y-f neighboring points is set as the value and stored in the image stacking unit 3 I '—' The stereo pixels of each stereo point are 315.) In the memory area (step, 200539052 V. invention description (6) (step 2 3 0); when all the pixels in the two-dimensional image data are converted into three-dimensional pixel values, they can be stacked into a three-dimensional image data (step 3 1 6) 〇 Please refer to "Figure 6" and "Figure 7", how to display the reconstructed 3D image File. “Figure 6” is a schematic diagram of the format of a 3D image file created based on the 3D image reconstruction method. As shown in the figure, the 3D image file 3 5 mainly contains header information 3 5 1 and s 1 ice image information 3 5 2. Among them, the header information 3 5 1 is used to describe the version announcement, copyright description, image content description, three-dimensional axis scale, volume size and image byte arrangement. The s 1 ice image information 3 5 2 is the first The image tribute from 1 to the Nth s 1 ice. "Figure 7" is a flow chart of 3D image display. When the user wants to confirm the image quality of the 3D image data reconstruction, the image display unit 3 2 Capture 3D image data (files) from the 3D image memory area in the image stacking unit 31 (step 3 2 1), and then analyze the header information of the 3D image data one by one (step 3 2 2) to confirm the header information Comply with the standard (step 3 2 3), and control the display mode of the image data according to the header information (step 3 2 4) to display the 3D image data according to the settings of the header information (step 3 2 5); this 3D image data It can be displayed by imaging technology such as 0pen GL. Please refer to "Figure 8" for the freehand 3D image distortion compensation method of the present invention. When the original scale of each axis (X, y, ζ) (sca 1 e ) Are (xsl, ysl, zsl) respectively, and then the axis scales are changed to (XS2, Υ S2, zS2) due to changes (for example, changes caused by free-hand scanning).

Page 11 200539052 V. Description of the invention (7) Uneven changes in the three-axis scale cause distortion and the number of pixels (0 1 and 6 1) is changed from (and (11, 7 (11, 2 (11) to ( And (12, 7 (12, 2 (12), then the number of pixels that should be on each axis is X dt = X dl * X S1 / X S2, y dt = y di * y si / ys and z dt = z dl * z S1 ./z si temple, you can use the distortion compensation method to perform the image distortion compensation process. The operation steps of the above-mentioned distortion compensation method are as follows: First, calculate the various items generated by the free hand scanning. The number of pixels on the axis (X dt, y dt, z (step 3 31); the number of pixels on each axis (x dt, y dt, z d1) is the pixel values relative to the X, Y, and Z axes after freehand scanning ( pixel).

Then, through the calculation formula md = max (x dt, y dt, z dt), confirm the number of maximum pixels after scanning (m J (step 3 3 2)). The number of maximum pixels (md) is after the scale adjustment (scanning). , The maximum number of pixels in each axis (eg, the number of pixels in each axis (X dt, y dt, z after scanning is (2 5 6 * 3 2 0 * 180), the maximum number of pixels ( It is 3 2 0). Then, confirm the number of maximum pixels (m J is not greater than 1M value (step 3 3 3); this M value is the largest of the pixel values (pixel) of the X, Υ, and Z axes before scanning) The number of pixels (pi xe 1). When the number of pixels of each axis is set to (2 5 6 * 2 5 6 * 2 5 6) before scanning, the M value is 2 5 6.

Then, according to the comparison result of step 3 3 3, obtain the scale adjustment value (sa) (step 3 3 4) and (step 3 3 5), and step 3 3 4 displays when the number of maximum pixels (mj is greater than the value of M, the scale adjustment The value (sa) can be obtained by the calculation formula of sa = M / m A. In step 3 3 5 it is displayed that when the number of maximum pixels (m J is not greater than M value, the scale adjustment value (s J is set 200539052) V. Description of the invention (8) Image z and then execute the adjustment procedure (step 3 3 6); the adjustment procedure * sa, ^ 2 = ^ and Zd2 = Zdt * s ^ The calculation formula depends on the number of primes (& , Yd2, Zd2). Then, get the new ys for each axis for each round; B, Zs3) (step 3 3 7); new scale / scale tJ for each axis? Xs3 = Xdl * Xsi / Xd2 '^ 3 = ydl * ysl / y ^ 2s3 = 2Xf I know the difference formula. dl Z si / 2: d2 ^ Finally, according to the new scale of each axis (h, the interpolation calculation of distortion compensation is obtained by the following formula: f = 于 料 ( Step 3 3 8): Pixel compensation data

For (X 2 1; Xd2; χ + +)

F〇r (y = i; y 仏; y ++)

For (z = i; zS xd2; z + +) i ^ Xdist = (xD * xs3 / xs1 + 0.5; Ydlst = (yD * ys3 / ys1 + 0.5; Z i (z 1) ^ zs3 / zs1 + 0.5; 〇 utdata3d [x] [y] [z] = i nda ta [Xdlst] [Ydist] [Zdist]} 'Xs 丨, J and J are the original scales of each axis, xs2, ^ 2 and zs2 are the f-axis after scanning The scale, xs3, ^ 3 and zs3 are the new scales after adjustment, and xd2, yd2 and zd2 are the number of new pixels on each axis after scanning) 丨 In step 2 0 4 above, if the number of pixels on each axis after adjustment is adjusted Not exceeding each axis

When pixels = number (xdt, ydt, zdt), the adjusted ratio will not be the same as the input image, but the three-axis ratio will be based on the true size to achieve the distortion compensation effect.

^ As mentioned above, the freehand 3D image reconstruction and distortion compensation method mentioned in the present invention can be applied not only to ultrasonic instruments, but also to digital X

Page 13 200539052 V. Description of the invention (9) Light (DX), computed tomography (CT), resonance imaging (MRI), endoscope, ultrasound (US), nuclear medicine (NM), In positron ray tomography (PET) and single-photon ray tomography (SPECT), the image distortion caused by the inconsistent scanning speed (for example, the uneven scale of each axis caused by the unsightly scanning of seedlings), provide The fast and accurate image compensation method enables the method of the present invention to quickly obtain two-dimensional image data of South resolution. The above are only the preferred embodiments of the present invention, and are not intended to limit the scope of implementation of the present invention; that is, all equivalent changes and modifications made in accordance with the scope of the patent application for the present invention are covered by the scope of the patent for the present invention .

I

Page 14 200539052 Brief description of the drawings. Figure 1 is a schematic diagram of the free-form 3D image reconstruction and distortion compensation method mentioned in the present invention. Figure 2 is a schematic diagram of the nearest neighbors of the three-dimensional points generated by interpolation. Figure 3 It is the operation flowchart of the freehand 3D image reconstruction method mentioned in the present invention; FIG. 4 is the operation flowchart of capturing images in the 3D image reconstruction method; and FIG. 5 is the operation flowchart of the direct stacking method in the 3D image reconstruction method Figure 6 is a schematic diagram of the format of a three-dimensional image file created by the three-dimensional image reconstruction method of the present invention; Figure 7 is a flowchart of a three-dimensional image display operation; and Figure 8 is a free-hand three-dimensional image distortion compensation method according to the present invention Operation flow chart. [Illustration of Symbols] 1 2D imaging system 2 2D image capture module 3 3D image reconstruction module 3 1 image stacking unit 3 2 image display unit 3 3 image distortion compensation unit 35 3D image file 351 files Head 3 5 2 s 1 ce image information 4 0 computer executable platform 5 0 digital image management system

Page 15 200539052 Simple illustration of 6C 丨 network step 200 Set a to determine the number of X and Y pixels in two-dimensional image data Step 201 Set the depth of the probe scan step 202 Set the scale of the X and Y axes Step 203 Set Capture a few seconds after the probe travels. Step 204 Press the image capture button. Step 205. Smoothly move the probe within the set time. -Irn. Δ and a fixed distance. Step 206. Store the 2D image in the digital image management system. Step 210. The number of two-dimensional pixels and the threshold determine the number of nearest neighbors. Step 220 Set the three-dimensional pixel values based on the two-dimensional pixel values of the nearest neighbor. Step 230 Stack the three-dimensional pixel values in order to become two-dimensional image data. Step 311 Image data interface capture two Dimensional pixel step 312 Store the two-dimensional image memory area Step 313 Find the nearest neighbor of the three-dimensional point Step 314 The three-dimensional pixel value is the two-dimensional pixel value of the nearest point Step 315 Store the three-dimensional pixel value in the two-dimensional image memory area Step 316 Rebuild completed Step 321 Two-dimensional image The display module obtains the two-dimensional image data. Step 322 analyzes the file data of the two-dimensional image. Step 323. Whether the header information meets the standard.

Page 16 200539052 Schematic description of steps 3 2 4 Step 3 2 5 Step 3 3 1 Step 3 3 2 Step 3 3 3 Step 334 Step 3 3 5 Step 3 3 6 Step 3 3 7 Step 3 3 8 Based on the profile information Manipulate the image data in the 3D image file to display the 3D image calculation of the number of pixels on each axis (xdt, ydt, zdt) generated by the free-hand scanning of the seedlings. Use the calculation formula md = max (xdt, ydt, zd1) to confirm the maximum after scanning. The number of pixels (m) confirm that the number of maximum pixels (in d) is not greater than one M value. Obtain a scale adjustment value according to the comparison result of step 3 3 3 (sj Obtain a scale adjustment value according to the comparison result of step 3 3 3 (sj perform adjustment The program obtains the new scales of each axis (X s3, y S3, Z S3). According to the new scales of each axis (x s3, y S3, z S3), the interpolation calculation applied to distortion compensation is to obtain three-dimensional pixel compensation data by the following formula. i

Page 17

Claims (1)

200539052 VI. Scope of patent application 1. A freehand 3D image reconstruction method for converting a 2D image data into a 3D image data, the method includes the following steps: continuously capturing the 2D image data; according to the 2D The plurality of two-dimensional pixel values and a threshold value in the image data determine the plurality of nearest neighbors of the plurality of three-dimensional points; the two-dimensional pixel values of the nearest neighbor are captured and set to the three-dimensional pixel value of the three-dimensional points (voxel ); And sequentially stacking the three-dimensional pixel values into the three-dimensional image data. 2. The free-hand 3D image reconstruction method described in item 1 of the scope of patent application, wherein the 2D image data is analog 2D image data. φ 3. The freehand 3D image reconstruction method described in item 1 of the scope of patent application, wherein the 2D image data is digital 2D image data. 4. The freehand 3D image reconstruction method as described in item 1 of the scope of patent application, wherein the 3D image data includes a file header information and an s 1 i c e image information. 5. The free-hand 3D image reconstruction method described in item 4 of the scope of the patent application, wherein the slot header tribute message includes version announcement, copyright notice, image content description, three-dimensional axis scale, volume size, and image byte arrangement the way. 6. —A freehand 3D image distortion compensation method, applied to 4D 3D image data acquired after freehand scanning, the method includes the following steps: selecting the number of pixels of each axis in the 3D image data (X dt, y dt , Z dt) is one of the maximum number of pixels (md); confirm the ratio of the number of maximum pixels (md) to a value of M, and generate Page 18 200539052 VI. One corresponding scale adjustment value (S a) of the scope of patent application; execute the adjustment procedure to adjust the scale adjustment value (S 3) and the number of pixels of each axis (X dt, y dt, z d1) The product of) yields the number of new pixels on each axis (X d2, y d2, z d2); according to the number of new pixels on each axis (x d2, y d2, z dZ) and the number of pixels on each axis (x The product of the ratio of dt, y dt, z and an original scale (x sl, y sl, z sl) produces a new scale (x s3, y s3, z s3); and according to the new scale (x s3, y s3) , Z s3) and the original scale (x si, y sl, z sl) use an interpolation algorithm to generate a three-dimensional pixel compensation data. 礞 | 7. The free-hand three-dimensional image described in item 6 of the scope of patent application Distortion compensation method, wherein the M value is the largest number of pixels selected from the pixel values of the X, Y, and Z axes before scanning. 8. Freehand 3D image distortion compensation as described in item 6 of the scope of patent application Method, wherein the number of maximum pixels (m is selected from the number of pixels of each axis after scanning (X dt, y dt z dt). 9. The freehand 3D image distortion compensation method described in item 6 of the patent application scope, wherein the M value is greater than the number m of the maximum pixels, and the scale adjustment value (s J It is 1 and no distortion compensation is performed. 1 0. The freehand 3D image distortion compensation f method described in item 6 of the scope of patent application, wherein when the M value is not greater than the number of maximum pixels (md), the scale adjustment value (S) can be obtained by the calculation formula of sa = M / mA. 1 1. The free-hand three-dimensional image distortion compensation method described in item 6 of the scope of patent application, wherein the new scale (XS3 'y * s3' Zs3) can be obtained by Xs3 = Xdl Page 19, 200539052 6. Scope of patent application * xsl / xd2, 753 = 7〇11 * 731/7 (1 and zs3 = zdl * zsl / zd2 ^ Calculate the formula to obtain 〇1 2 The free-hand three-dimensional image distortion compensation method described above, wherein the number of new pixels on each axis (X d2, y d2, z d2) can be calculated by using xd2 = xdt * sa, yd2 = ydt * s, and zd2 = zdt * s. The calculation formula is obtained. 1 3. The freehand 3D image distortion compensation method described in item 6 of the scope of patent application, wherein the 3D pixel compensation data is calculated through the following formula: For (x = l; x 'X d2v x + +) For (y = 1; y 仏; y + +) For (z = 1; z d2; z + +) {{X dis 卜 （x-1) * xs 3 / XS 1 + 0 · 5 ; Yd ist = (y-1) * ys 3 / ys 1 + 0 _ 5; Zdist = (z-1) 氺 zs3 / zs1 + 0.5; out data3d [x] [y] [z] = i ndata [Xd ist] [Yd ist] [Zd ist]} where xsbu ”1 and ZS1 are the original scales of each axis, xs2, 782 and zs2 are the scales of each axis after scanning the seedlings, xs3, yS3 and 3 are the The new scale, xd2, yd2 and zd2 are the number of new pixels on each axis after scanning. I Page 20