Systems and methods for generating three-dimensional images

Abstract

This invention provides a system and method for generating three-dimensional images, comprising: capturing a set of color images of an object, the set of color images being related to color information of the object; generating a first point cloud based on at least the set of color images, the first point cloud being related to a rigid surface and a non-rigid surface of the object; performing a filtering operation on the first point cloud based on the set of color images to generate a second point cloud, the second point cloud having a higher correlation with the rigid surface than with the non-rigid surface; selectively performing a matching operation using the second point cloud and a target point cloud to obtain pose information; and combining the first point cloud with the target point cloud based on the pose information to update the target point cloud, thereby generating a three-dimensional image of the object. The system and method for generating three-dimensional images provided by this invention can reduce the interference of O2 from non-rigid surfaces, thereby generating pose information, and thereby integrating newly captured images into the target image to improve the quality of the generated three-dimensional image.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of oral scanning, and in particular, to a system and method for generating a three-dimensional image. BACKGROUND

[0002] With the advancement of technology, more and more professionals are willing to use optical auxiliary devices to improve the convenience and accuracy of operation. For example, in the field of dentistry, there are currently intraoral scanners to assist dentists in detecting the oral cavity. Intraoral scanners can capture images of the oral cavity and convert them into digital data to help professionals such as dentists and denture technicians in diagnosis and treatment and denture making.

[0003] However, in actual operation, it is found that when using an intraoral scanner to obtain images of teeth, images of gums and tongues and the like are also obtained, and these non-tooth images often cause interference, resulting in a decrease in the quality of the captured tooth images. In addition, interference often leads to the failure of newly captured images to successfully merge into the generated three-dimensional image, resulting in a decrease in operation speed.

[0004] Therefore, it is necessary to design a new system and method for generating a three-dimensional image to overcome the above-mentioned defects. SUMMARY

[0005] The present application aims to provide a system and method for generating a three-dimensional image, which can effectively improve the quality of tooth images.

[0006] To achieve the above-mentioned purpose, the present application provides a method for generating a three-dimensional image, which comprises: capturing a set of color images of an object, the set of color images being related to color information of the object; generating a first point cloud according to at least the set of color images, the first point cloud being related to a rigid surface and a non-rigid surface of the object; performing a screening operation on the first point cloud according to the set of color images to generate a second point cloud, the second point cloud having a higher correlation with the rigid surface than with the non-rigid surface; selectively performing a matching operation using the second point cloud and a target point cloud to obtain pose information; and combining the first point cloud to the target point cloud according to the pose information to update the target point cloud, to generate a three-dimensional image of the object.

[0007] Preferably, the second point cloud is related to the rigid surface and not related to the non-rigid surface.

[0008] Preferably, the method further comprises: capturing a set of structure images of the object, the set of structure images being related to depth information of the object; and generating the first point cloud according to the set of color images, comprising: generating the first point cloud according to the set of color images and the set of structure images.

[0009] Preferably, the method further comprises: capturing the set of structure images of the object, including: projecting a set of light patterns to the object; and generating the set of structure images according to distribution of the set of light patterns on a surface of the object; generating the first point cloud according to the set of color images and the set of structure images, including: processing the set of structure images to obtain depth information of the object; obtaining color information of the object according to the set of color images; and generating the first point cloud according to the depth information and the color information.

[0010] Preferably, the performing the filtering operation on the first point cloud according to the set of color images to generate the second point cloud comprises: processing the set of color images to obtain color information of the object; generating a mask according to the color information; and processing the first point cloud using the mask to generate the second point cloud.

[0011] Preferably, the method further comprises: the color information of the object comprises red-green-blue color information; the generating the mask according to the color information comprises: converting the color information to color space information; processing the color space information using an upper limit value and a lower limit value to select partial data points in the first point cloud; and generating the mask according to the selected partial data points.

[0012] Preferably, the target point cloud is related to the rigid surface and the non-rigid surface; or the target point cloud is related to the rigid surface and not related to the non-rigid surface.

[0013] Preferably, the method further comprises: capturing a new point cloud of the object to generate a new point cloud; and combining the new point cloud to a part of the target point cloud to update the target point cloud; wherein the new point cloud is related to the rigid surface and the non-rigid surface, and the part of the target point cloud is related to the rigid surface and not related to the non-rigid surface.

[0014] Preferably, the matching operation comprises: obtaining a set of rigid surface data points according to the second point cloud and the target point cloud; obtaining a set of non-rigid surface data points according to the second point cloud and the target point cloud; multiplying a number of the set of rigid surface data points by a first weight value to obtain a first product; multiplying a number of the set of non-rigid surface data points by a second weight value to obtain a second product; and dividing a sum of the first product and the second product by a sum of the first weight value and the second weight value to obtain a value; if the value is less than a threshold value, discarding the second point cloud.

[0015] Preferably, the matching operation comprises an iterative closest point operation.

[0016] Based on the above method of generating a three-dimensional image, the present application also provides a system for generating a three-dimensional image, which comprises: a scanner for capturing a set of color images of an object, the set of color images being related to color information of the object; a processor for generating a first point cloud based on at least the set of color images, performing a screening operation on the first point cloud based on the set of color images to generate a second point cloud, selectively performing a matching operation using the second point cloud and a target point cloud to obtain pose information, and combining the first point cloud to the target point cloud based on the pose information to update the target point cloud to generate a three-dimensional image of the object; and a display for displaying the three-dimensional image of the object; wherein the first point cloud is related to a rigid surface and a non-rigid surface of the object, and the correlation of the second point cloud with the rigid surface is higher than the correlation of the second point cloud with the non-rigid surface.

[0017] Preferably, the system further comprises: the scanner is also used to capture a set of structure images of the object, the structure images being related to depth information of the object; and the processor generates the first point cloud based on the set of color images and the set of structure images.

[0018] Preferably, the scanner is an oral cavity scanner, the object is an oral cavity, the rigid surface is a tooth surface in the oral cavity, and the non-rigid surface is a non-tooth surface in the oral cavity.

[0019] Compared with the prior art, the system and method for generating a three-dimensional image provided by the present application can reduce the interference of the non-rigid surface O2, thereby generating pose information and integrating the newly captured image into the target image to improve the quality of the generated three-dimensional image. For example, when scanning the oral cavity, the interference of the gums and tongue can be reduced to improve the modeling quality of the three-dimensional image of the teeth. Since the unexpected interference can be reduced, the probability of point cloud matching failure leading to forced interruption or image fusion failure can be reduced, thereby improving the smoothness of the scanning operation and shortening the time of the scanning operation. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 Structure schematic diagram of the system for generating a three-dimensional image provided by an embodiment of the present application;

[0021] Figure 2 Flowchart of the method for generating a three-dimensional image provided by an embodiment of the present application;

[0022] Figure 3 Flowchart of capturing a structure image and generating a first point cloud based on the structure image and the color image provided by an embodiment of the present application;

[0023] Figure 4 Flowchart of performing a screening operation on a first point cloud based on a color image to generate a second point cloud provided by an embodiment of the present application;

[0024] Figure 5 A schematic diagram of the process for masking the first point cloud provided in an embodiment of the present invention;

[0025] Figure 6 This is a flowchart illustrating the process of determining whether to use a second point cloud during a matching operation, as provided in an embodiment of the present invention. Detailed Implementation

[0026] To provide a better understanding of the purpose, structure, features and functions of the present invention, detailed descriptions are provided below with reference to specific embodiments.

[0027] Certain terms are used in the specification and claims to refer to specific elements. It will be understood by those skilled in the art that manufacturers may use different names to refer to the same element. This specification and claims do not distinguish elements by differences in name, but rather by differences in function. The term "comprising" throughout the specification and claims is an open-ended term and should be interpreted as "comprising but not limited to".

[0028] Please see Figures 1 to 6 , Figure 1 This is a schematic diagram of the structure of a system for generating three-dimensional images provided in an embodiment of the present invention. Figure 2 This is a flowchart illustrating the method for generating three-dimensional images provided in an embodiment of the present invention. Figure 3 This is a schematic diagram illustrating the process of capturing structural images and generating a first point cloud based on structural and color images, as provided in an embodiment of the present invention. Figure 4 This is a schematic diagram illustrating the process of performing a filtering operation on a first point cloud based on a color image to generate a second point cloud, provided by an embodiment of the present invention. Figure 5 This is a schematic diagram of the process for masking the first point cloud according to an embodiment of the present invention. Figure 6 This is a flowchart illustrating the process of determining whether to use a second point cloud during a matching operation, as provided in an embodiment of the present invention.

[0029] To avoid interference and improve the quality of the captured images, embodiments can provide solutions as described below. Figure 1 This is a schematic diagram of the system 10 for generating three-dimensional images in an embodiment. Figure 1 As shown, system 100 may include scanner 102, processor 104, and display 106. Scanner 102 can be used to scan object O, processor 104 can perform related processing, and display 106 can display a three-dimensional image of object O. For example, scanner 102 may be a dental scanner, and object O may be a dental cavity, such as... Figure 1 As shown, an object O may include a rigid surface O1 (e.g., teeth) and a non-rigid surface O2 (e.g., gums, tongue, etc.).

[0030] likeFigure 1 As shown, when scanning object O to generate a three-dimensional image, light ray L1 can be projected onto object O, and the reflected light ray L2 can be received to capture the image.

[0031] According to an embodiment, the light beam L1 may contain a specific structured light pattern, such as a striped pattern or a checkered pattern. The scanner 102 captures an image of the scanned object O, and the distribution of the light pattern on the received image is decoded to obtain a point cloud. Each point cloud can be a set of data points in space, which can be used to represent the three-dimensional shape of the object O. Taking oral scanning as an example, the user can move the scanner 102 (e.g., a dentist can hold the oral scanner and move it gradually) to scan multiple areas within the oral cavity, thereby capturing multiple images. These multiple images are then stitched together to produce a more complete three-dimensional image. According to an embodiment, the interference from the non-rigid surface O2 can also be reduced based on the color information of the object O to better construct the three-dimensional image of the object O, as described below.

[0032] Figure 2 for Figure 1 The flowchart below illustrates a method 100 for generating a 3D image of object O. Method 100 may include the following steps:

[0033] Step 110: Capture a set of color images of object O, wherein the set of color images is related to the color information of object O;

[0034] Step 120: Generate a first point cloud P based on at least the set of color images (e.g., Figure 5 As shown), the first point cloud P can be related to the rigid surface O1 and the non-rigid surface O2 of the object O.

[0035] Step 130: Perform a filtering operation on the first point cloud P based on the set of color images to generate the second point cloud P' (e.g., Figure 5 As shown in the figure, the correlation between the second point cloud P' and the rigid surface O1 is higher than the correlation between the second point cloud P' and the non-rigid surface O2.

[0036] Step 140: Selectively perform a matching operation using the second point cloud P' and the target point cloud to obtain pose information; and

[0037] Step 150: Combine the first point cloud P with the target point cloud based on the pose information to update the target point cloud and generate a 3D image of object O.

[0038] According to the embodiments, such as Figure 1 and Figure 2 As shown, the scanner 102 can execute step 110, the processor 104 can execute steps 120 to 150, and the display 106 can display the three-dimensional image generated in step 150.

[0039] The following description uses an oral cavity scan as an example to illustrate method 100. In method 100, object O can be the area surrounding the teeth. Therefore, object O can include a rigid surface O1 (e.g., teeth) and a non-rigid surface O2 (e.g., non-tooth areas such as the gums, tongue, etc.).

[0040] In step 110, the set of color images may contain multiple images, which can be obtained by sequentially projecting light of various colors (such as red, green, and blue) or various wavelengths (such as visible light, infrared light, and ultraviolet light) onto object O and then capturing the images. Therefore, the color information of object O can be obtained through this set of color images.

[0041] In step 120, the resulting first point cloud P can represent the shape of the rigid surface O1 and the non-rigid surface O2 of object O in three-dimensional space.

[0042] According to an embodiment, since the rigid surface O1 and the non-rigid surface O2 reflect and absorb light to different degrees, step 130 can be performed accordingly. In step 130, a filtering operation can be performed on the first point cloud P based on the color information of the object O to generate a second point cloud P'. The correlation between the second point cloud P' and the rigid surface O1 (e.g., the tooth surface) can be higher than its correlation with the non-rigid surface O2 (e.g., the gum surface). According to an embodiment, in another case, the second point cloud P' can be correlated only with the rigid surface O1 and not with the non-rigid surface O2. In other words, compared to the first point cloud P, the proportion of data from the rigid surface O1 is higher in the second point cloud P', and the proportion of data from the non-rigid surface O2 has been filtered out and is therefore lower.

[0043] In step 140, for example, the matching operation may include (but is not limited to) an Iterative Closest Points (ICP) operation. The target point cloud may be a result point cloud fused from previously generated point clouds. For example, when an intraoral scanner is moved to capture images of multiple parts of the oral cavity, multiple point clouds can be generated from these images, and these point clouds can be sequentially combined to produce a target point cloud. Therefore, the target point cloud may correspond to a relatively large and complete three-dimensional image of object O.

[0044] In step 140, if the proportion of matchable points between the second point cloud P' and the target point cloud is too low, the matching operation can be abandoned. Therefore, the matching operation can be selectively performed. For example, if the dentist moves the dental scanner too fast, causing the corresponding areas of the second point cloud P' and the target point cloud to be too far apart, the proportion of matchable points will be too low, and the second point cloud P' will not be combined with the target point cloud.

[0045] According to the embodiment, the pose information mentioned in step 140 may include a pose matrix, a rotation matrix, and / or a translation matrix, and the pose information may be a positional relationship function between the newly entered second point cloud P' and the target point cloud.

[0046] According to the embodiment, the pose information is not generated based on the first point cloud P and the target point cloud, but rather based on the second point cloud P' and the target point cloud. Since the pose information is generated based on the second point cloud P', and the second point cloud P' has a high correlation with the rigid surface O1, by using the pose information combined with the new point cloud and the target point cloud, the influence and interference caused by the information of the non-rigid surface O2 can be reduced.

[0047] In step 150, the pose information is used to combine the first point cloud P with the target point cloud, which can avoid interference caused by the non-rigid surface O2, thereby improving the scanning smoothness and the quality of the stitched 3D image.

[0048] like Figure 1 As shown, according to an embodiment, a set of structural images of object O can be separately captured, wherein the set of structural images can be correlated with the depth information of object O. According to an embodiment, in step 120, a first point cloud P can be generated based on the set of color images and the set of structural images. Figure 3 As an example, this is a flowchart of capturing a structural image and generating a first point cloud P based on the structural image and color image. Figure 3 As shown, it may include the following steps:

[0049] Step 310: Project a set of light patterns onto object O;

[0050] Step 320: Generate the set of structural images based on the distribution of the set of light patterns on the surface of object O;

[0051] Step 330: Process the set of structural images to obtain the depth information of object O;

[0052] Step 340: Obtain the color information of object O based on the set of color images; and

[0053] Step 350: Generate the first point cloud P based on depth and color information.

[0054] According to an embodiment, the set of light patterns described in step 310 may include specific patterns, such as stripes and checkerboard patterns. For example, the set of light patterns in step 310 may include multiple light patterns with different stripe widths and stripe spacings.

[0055] In step 320, a light pattern can be projected onto object O, and the distribution of the light pattern on object O can be photographed to generate a structural image, thereby generating depth information in step 330.

[0056] Step 340 can correspond to Figure 2 In step 110, and in step 340, a first point cloud P can be generated based on the depth information and color information of object O. The depth information pertains to the three-dimensional structure of object O, and the color information pertains to the color of object O.

[0057] According to an embodiment, Figure 1 The scanner 102 can be used to perform steps 310 and 320, and to capture the structural image of object O. Figure 1 The processor 104 in the middle can be used to execute steps 330 to 350.

[0058] Figure 4 for Figure 2 The diagram illustrates how a filtering operation is performed on the first point cloud P based on the color image to generate the second point cloud P'. In other words, Figure 4 Can be Figure 2 The flowchart for step 130. Figure 5 for Figure 4 The diagram illustrates the process of processing the first point cloud P using a mask M. (See diagram for example.) Figure 4 As shown, the following steps can be performed:

[0059] Step 410: Process the color image to obtain the color information of object O;

[0060] Step 420: Generate mask M based on color information; and

[0061] Step 430: Process the first point cloud P using mask M to generate the second point cloud P'.

[0062] According to an embodiment, the color information in step 410 can be red, green, and blue (RGB) color information. In step 420, according to an embodiment, the color information can be converted into color space information. For example, Gaussian blur processing can be performed on the color information, followed by color space conversion, to convert the red, green, and blue (RGB) color information into color space information.

[0063] According to an embodiment, color space information may include hue-saturation-brightness (HSV) information, hue-saturation-brightness (HSL) information, and / or brightness-chroma (YUV) information.

[0064] Then, upper and lower limits can be used to process color space information (e.g., HSV, HSL, or YUV information) to select a subset of data points P1 in the first point cloud P (e.g., data points corresponding to teeth but not gums), and a mask M is generated based on the selected data points P1. The number of points in the first point cloud P is greater than or equal to the number of points in the selected data points P1.

[0065] For example, if the color space information is HSV information, the corresponding value of each point in the first point cloud P in the HSV space can be obtained (e.g., between 0 and 255). Therefore, based on predetermined upper and lower limits, data points of non-rigid surfaces O2 (e.g., gums and tongue) can be filtered out, and the remaining data points P1 can correspond to rigid surfaces O1 (e.g., teeth).

[0066] Since red, green, and blue (RGB) color information is easily affected by the intensity of light, converting color information into color space information can improve the quality of the resulting images.

[0067] In step 420, binarization can be performed based on color space information to generate a mask M. Binarization can classify colors into only two categories, such as black and white. Therefore, after binarization, the mask M can be generated using the area determined to be a non-rigid surface O2.

[0068] like Figure 5 As shown, the mask M generated in step 420 can correspond to the non-rigid surface O2 of object O (e.g., the gingival portion). Therefore, in step 430, by using the mask M to mask the first point cloud P, data points corresponding to the non-rigid surface O2 can be filtered out, thereby generating a second point cloud P' that is more correlated with the rigid surface O1.

[0069] According to an embodiment, a light ray L1 with a predetermined structured light pattern (e.g., a striped light pattern) can be projected onto an object O, and the light ray L2 can be received to generate an image. The generated image can then be covered by the aforementioned mask M. Afterwards, new depth information can be generated through decoding operations. Then, pose information (e.g., a pose matrix) can be obtained using a combination operation (e.g., an iterative nearest-point operation), and the pose information can be used to combine the first point cloud P into the target point cloud to update the target point cloud.

[0070] According to an embodiment, Figure 2 The target point cloud described in steps 140 and 150 can be related to the rigid surface O1 and the non-rigid surface O2 of the object O. For example, the target point cloud can include data points of the tooth surface and the gingival surface.

[0071] According to another embodiment, the target point cloud may be correlated with a rigid surface O1 but not with a non-rigid surface O2. For example, the target point cloud may include data points on the tooth surface but not data points on the gingival surface.

[0072] According to another embodiment, a set of recent images of object O can be captured to generate a recent point cloud. Then, the recent point cloud can be combined with a portion of a target point cloud (denoted as Q1) to update the target point cloud. The recent point cloud can be correlated with both the rigid surface O1 and the non-rigid surface O2 of object O, and the portion Q1 of the target point cloud can be correlated with the rigid surface O1 of object O but not with the non-rigid surface O2. For example, it can be used... Figure 1 The scanner 102 scans the object O to generate a new image and thereby generate a new point cloud, which may include information about teeth and gums; however, the part Q1 of the target point cloud used to combine with the new point cloud may only contain information about teeth.

[0073] Figure 6 Can be Figure 2 The flowchart for determining whether to use the second point cloud P' in the matching operation of step 140.

[0074] like Figure 6 As shown, the following steps can be performed:

[0075] Step 610: Obtain a set of rigid body surface data points (denoted as P'1) based on the second point cloud P' and the target point cloud;

[0076] Step 620: Obtain a set of non-rigid surface data points (denoted as P'2) based on the second point cloud P' and the target point cloud;

[0077] Step 630: Multiply the number of rigid body surface data points P'1 in this set by the first weight (denoted as W1) to obtain the first product;

[0078] Step 640: Multiply the number of non-rigid surface data points P'2 in this set by the second weight (denoted as W2) to obtain the second product;

[0079] Step 650: Divide the sum of the first product and the second product by the sum of the first weight W1 and the second weight W2 to obtain the value V;

[0080] Step 660: Determine if the value V is less than the threshold value; if yes, proceed to step 670; if no, proceed to step 680.

[0081] Step 670: If the value V is less than the threshold value, discard the second point cloud P'; and

[0082] Step 680: If the value V is not less than the threshold value, use the second point cloud P'.

[0083] Steps 630 to 650 can be expressed as the following formula:

[0084] V=[(P'1×W1)+(P'2×W2)] / (W1+W2);

[0085] For example, the first weight W1 (e.g., the weight of the tooth portion) could be 0.8, and the second weight W2 (e.g., the weight of the non-tooth portion) could be 0.2. Or, for example, the first weight W1 could be 0.5, and the second weight W2 could be 0.5. By processing the data in a weighted manner, data on non-rigid surface O2 can be partially excluded.

[0086] Figure 6 In the process, data from non-rigid surface O2 may not be completely excluded; instead, data from rigid surface O1 and non-rigid surface O2 may be used in a weighted manner. In step 670, if the matching fails, the second point cloud P' may be discarded. In step 680, if the matching succeeds, the second point cloud P' may be used to generate the aforementioned pose information (e.g., a pose matrix), which is then used to integrate the newly generated point cloud into the target point cloud and, based on this, to integrate the newly acquired image into the target image.

[0087] According to the embodiment, if most of the data points of the newly captured image belong to non-rigid surfaces O2, for example, 80% of the data points belong to non-tooth surfaces, then the image can be dynamically adjusted to enter the non-rigid surface stitching mode, or enter the weak stitching mode that prioritizes smoothness, so as to moderately sacrifice the accuracy of the area, but improve the smoothness of the scanning operation.

[0088] In summary, the system and method provided by the embodiments of the present invention can reduce interference from O2 on non-rigid surfaces, thereby generating pose information and integrating newly acquired images into target images to improve the quality of the generated 3D images. For example, when scanning the oral cavity, interference from the gums and tongue can be reduced, thus improving the modeling quality of the 3D images of the teeth. Because unexpected interference can be reduced, the probability of point cloud matching failure leading to forced interruption or image fusion failure can be reduced, thereby improving the smoothness of the scanning operation and shortening the scanning time. Therefore, the present invention is helpful in addressing the problems in this field.

[0089] The present invention has been described in the above-described embodiments; however, these embodiments are merely examples for implementing the present invention. It must be noted that the disclosed embodiments do not limit the scope of the present invention. Conversely, any modifications and refinements made without departing from the spirit and scope of the present invention are within the scope of patent protection of the present invention.

Claims

1. A method of producing a three-dimensional image, characterized by, The method comprises: capturing a set of color images of an object, the set of color images being related to color information of the object; generating a first point cloud from at least the set of color images, the first point cloud being related to rigid surfaces and non-rigid surfaces of the object; performing a filtering operation on the first point cloud from the set of color images to generate a second point cloud, the second point cloud having a higher relevance to the rigid surfaces than to the non-rigid surfaces; selectively performing a matching operation using the second point cloud and a target point cloud to obtain pose information; and combining the first point cloud to the target point cloud according to the pose information to update the target point cloud to generate a three-dimensional image of the object; wherein the matching operation comprises: obtaining a set of rigid surface data points from the second point cloud and the target point cloud; obtaining a set of non-rigid surface data points from the second point cloud and the target point cloud; multiplying a number of the set of rigid surface data points by a first weight to obtain a first product; multiplying a number of the set of non-rigid surface data points by a second weight to obtain a second product; and dividing a sum of the first product and the second product by a sum of the first weight and the second weight to obtain a value; and discarding the second point cloud if the value is less than a threshold value.

2. The method of generating a three-dimensional image of claim 1, wherein, The second point cloud is related to the rigid surfaces and not related to the non-rigid surfaces.

3. The method of generating a three-dimensional image of claim 1, wherein, The method further comprises: capturing a set of structure images of the object, the set of structure images being related to depth information of the object; and generating the first point cloud from the set of color images comprises generating the first point cloud from the set of color images and the set of structure images.

4. The method of generating a three-dimensional image of claim 3, wherein, The method specifically comprises: capturing the set of structure images of the object comprises projecting a set of light patterns to the object; and generating the set of structure images from a distribution of the set of light patterns on a surface of the object; generating the first point cloud from the set of color images and the set of structure images comprises processing the set of structure images to obtain depth information of the object; obtaining color information of the object from the set of color images; and generating the first point cloud from the depth information and the color information.

5. The method of generating a three-dimensional image of claim 1, wherein, The performing the filtering operation on the first point cloud from the set of color images to generate the second point cloud specifically comprises: processing the set of color images to obtain color information of the object; generating a mask from the color information; and processing the first point cloud using the mask to generate the second point cloud.

6. The method of generating a three-dimensional image according to claim 5, wherein, The method further comprises: The color information of the object comprises red-green-blue color information; The generating the mask from the color information specifically comprises converting the color information to color space information; processing the color space information using an upper limit value and a lower limit value to select a portion of data points in the first point cloud; and generating the mask from the selected portion of data points.

7. The method of generating a three-dimensional image of claim 1, wherein, The target point cloud is related to the rigid surfaces and the non-rigid surfaces; or the target point cloud is related to the rigid surfaces and not related to the non-rigid surfaces.

8. The method of generating a three-dimensional image of claim 1, wherein, The method further comprises: capturing a set of new progress images of the object to generate a new progress point cloud; and combining the new progress point cloud to a portion of the target point cloud to update the target point cloud. wherein the new incoming point cloud is related to the rigid surface and the non-rigid surface, and the portion of the target point cloud is related to the rigid surface and not related to the non-rigid surface.

9. The method of generating a three-dimensional image of claim 1, wherein, The matching operation includes an iterative closest point operation.

10. A system for producing a three-dimensional image, characterized by The system includes: a scanner to capture a set of color images of an object, the set of color images being related to color information of the object; a processor to generate a first point cloud from at least the set of color images, perform a screening operation on the first point cloud from the set of color images to generate a second point cloud, selectively perform a matching operation using the second point cloud and a target point cloud to obtain pose information, wherein the matching operation includes: obtaining a set of rigid surface data points from the second point cloud and the target point cloud; obtaining a set of non-rigid surface data points from the second point cloud and the target point cloud; multiplying a number of the set of rigid surface data points by a first weight to obtain a first product; multiplying a number of the set of non-rigid surface data points by a second weight to obtain a second product; and dividing a sum of the first product and the second product by a sum of the first weight and the second weight to obtain a value; if the value is less than a threshold value, discarding the second point cloud; and, incorporating the first point cloud to the target point cloud according to the pose information to update the target point cloud to generate a three-dimensional image of the object; and a display to display the three-dimensional image of the object. wherein the first point cloud is related to a rigid surface and a non-rigid surface of the object, and the second point cloud has a higher correlation with the rigid surface than with the non-rigid surface.

11. The system for generating a three-dimensional image of claim 10, wherein, The system further includes: The scanner is further to capture a set of structure images of the object, the structure images being related to depth information of the object; and The processor to generate the first point cloud from the set of color images and the set of structure images.

12. The system for generating a three-dimensional image of claim 10, wherein, The scanner is an oral scanner, the object is an oral cavity, the rigid surface is a tooth surface in the oral cavity, and the non-rigid surface is a non-tooth surface in the oral cavity.

Images