TWI623912B - Depth-image constructed method and system using the same - Google Patents

Abstract

A method for constructing a depth image includes acquiring a first planar image and a second planar image of a three-dimensional object at a starting position on a plane and a control position rotated by a preset angle from the center of the circle on the plane relative to the starting position. The first and second plane images have a bottom line respectively; the second plane image is rotated according to a preset angle so that the bottom lines of the first and second plane images are parallel to each other; the first plane image is translated in a direction perpendicular to the bottom line of the second plane image and At least one of the second plane images aligns the bottom line of the second plane image with the bottom line of the first plane image; and after translation, calculates and records the distance of the three-dimensional object relative to the plane according to the first plane image and the second plane image, and records Corresponding pixels in the depth image to generate a depth image.

Description

Deep image construction method and its application system

The invention relates to a method for constructing a depth image, and more particularly to a system for applying a method for constructing a depth image.

Traditional depth image construction methods and systems usually require multiple image capture devices to simultaneously illuminate a scene with depth to obtain the depth information of the scene and three-dimensional objects in the scene. In order for multiple image capture devices to illuminate a scene with depth at the same time, it takes a lot of time to calibrate the optical axis, adjust the light field, etc., to obtain the prerequisites. Therefore, in addition to the application of traditional depth image construction methods and systems, in addition to higher construction costs, more time must be spent on pre-operations and subsequent related calculations in order to obtain a scene and three-dimensional objects in the scene. In depth information. These harsh conditions further restrict the development of traditional depth image construction methods and systems, and hinder the popularization of related depth image applications. It can be seen that the above existing architecture obviously still has inconveniences and defects, and needs to be further improved. In order to solve the above-mentioned problems, the related fields have made great efforts to seek Seeking a solution, but for a long time no suitable method has been developed and completed. Therefore, how to effectively solve the above problems is really one of the important R & D topics at present, and it has become an urgent target for improvement in related fields.

One technical aspect of the present invention relates to a method for constructing a depth image, which captures multiple planar images of a three-dimensional object correspondingly at different positions on a plane, and uses at least two of the plurality of planar images to calculate the relative plane of the three-dimensional object Distance to produce a depth image.

The invention provides a depth image construction method for generating a first depth image according to a three-dimensional object. The first depth image may include a plurality of depth pixels. The depth image construction method includes (a) capturing a first planar image of a three-dimensional object at a starting position on a plane, wherein the first planar image has a first image baseline; (b) rotating from the starting position relative to a center of a circle on the plane Preset the angle to the control position, and capture a second plane image of the three-dimensional object at the control position, where the second plane image has the second image baseline; (c) rotate the second plane image according to the preset angle to make the first image baseline Parallel to the second image bottom line; (d) translating at least one of the first plane image and the second plane image in a direction perpendicular to the second image base line, so that the second image bottom line of the second plane image and the first plane image are Align the bottom line of the first image; and (e) calculate the distance of the three-dimensional object relative to the plane from the first plane image and the second plane image after translation, and record the corresponding depth pixel in the depth pixel to generate a first depth image .

In one or more embodiments of the present invention, the step (c) may further include (c1) rotating the first plane image according to the correction angle, so that the first image is The line and the horizontal line of the three-dimensional object are parallel to each other; and (c2) rotating the second plane image according to the correction angle and the preset angle, so that the first image bottom line and the second image bottom line are parallel to each other.

In one or more embodiments of the present invention, the normal of the plane passing through the center of the circle intersects the three-dimensional object at least at a certain point.

In one or more embodiments of the present invention, the above-mentioned fixed point lies outside the plane.

In one or more embodiments of the present invention, each of the first and second planar images includes a plurality of planar pixels. Step (d) may further include (d1) calculating the correspondence between the planar pixels of the first planar image and the planar pixels of the second planar image and the length in space; and (d2) calculating and along the vertical first object according to the correspondence. The direction of the bottom line of the image translates at least one of the first planar image and the second planar image.

In one or more embodiments of the present invention, the above-mentioned depth image construction method further includes changing the starting position, and repeating steps (a) to (e) to generate another depth image; and converting another depth image; Each depth pixel in is averaged with a corresponding depth pixel in the depth image to update the depth pixel of the depth image to generate a second depth image.

The invention provides a depth image construction system including a rotatable mechanism, an image capturing device, and a computing module. The rotatable mechanism is configured to rotate in a plane relative to the center of the circle. The image capturing device is arranged on the rotatable mechanism and is a distance from the center of the circle. When the rotatable mechanism rotates at a plurality of positions on the plane relative to the center of the circle to drive the image capturing device, the image capturing device is configured to capture a plurality of planar images of the three-dimensional object at the plurality of positions, respectively. Multiple locations can form a reference circle. The plane does not pass through a three-dimensional object. The computing module may include a first programming. The first programming is configured to generate a first depth image according to at least two of the planar images.

In one or more embodiments of the present invention, the normal of the image capturing device is parallel to the normal of the plane.

In one or more embodiments of the present invention, the plurality of positions corresponding to the plurality of plane images may include a start position and a reference position. A center angle is formed along the reference circle between the starting position and the reference position, and the first programming further uses the distance and the center angle to generate a first depth image.

In one or more embodiments of the present invention, the above-mentioned depth image includes a plurality of depth pixels. The computing module may further include a second programming. The second programming is configured to generate a second depth image by averaging corresponding depth pixels in the plurality of first depth images according to the plurality of first depth images generated by pairing the plurality of plane images in pairs.

100‧‧‧Depth image construction system

200 / 200’‧‧‧ rotatable mechanism

220‧‧‧ cantilever

240‧‧‧ rotating bearing

260‧‧‧ track

280‧‧‧ Slider

300‧‧‧Image capture device

400‧‧‧ Computing Module

420‧‧‧first programming

440‧‧‧Second programming

500‧‧‧Three-dimensional object

520‧‧‧Horizontal

600A ~ 600E‧‧‧Planar image

600B ’/ 600B” / 600C ’/ 600C” ‧‧‧Planar image

620B / 620B ’/ 620C / 620C’‧‧‧ flat pixels

640A ~ 640E‧‧‧object plane image

660B‧‧‧The first image bottom line

660C‧‧‧Second Image Bottom Line

700‧‧‧Depth image construction method

800‧‧‧ first depth image

820‧‧‧ Depth Pixel

822‧‧‧ Depth Pixel

824‧‧‧ Depth Pixel

840‧‧‧ Object Depth Image

900‧‧‧ Second depth image

920‧‧‧ Depth pixels

A1 / A2 / A3 / A4‧‧‧Location

B1 / B2 / B3 / B4 / B5‧‧‧Location

C1 / C2 / C3‧‧‧ Center

D‧‧‧ direction

k‧‧‧first angle

L1 / L2‧‧‧ extension cable

P1 / P2 / P3‧‧‧Plane

r‧‧‧ distance

RA‧‧‧Rotary shaft

S710 ~ S760‧‧‧step

VL1‧‧‧ vertical line

X‧‧‧ fixed point

Φ‧‧‧Preset angle

In order to make the above and other objects, features, advantages, and embodiments of the present invention more comprehensible, the description of the drawings is as follows: FIG. 1 is a schematic diagram of a depth image construction system according to various embodiments of the present invention.

FIG. 2 is a schematic diagram of a depth image construction system according to another embodiment of the present invention.

FIG. 3 is a simple schematic diagram of a depth image construction system applied to a three-dimensional object in the real world according to various embodiments of the present invention.

FIG. 4A to FIG. 4E are respectively schematic diagrams of plane images captured by the depth image construction system according to various embodiments of the present invention at different positions.

FIG. 5 is a flowchart illustrating a method for constructing a depth image according to various embodiments of the present invention.

6A to 6D are simple schematic diagrams of different steps in an imaging space in a method for constructing a depth image according to various embodiments of the present invention.

FIG. 7 is a schematic diagram of a depth image according to various embodiments of the present invention.

FIG. 8 is a simple schematic diagram of different steps in an imaging space in a method for constructing a depth image according to another embodiment of the present invention.

FIG. 9 is a schematic diagram of a depth image according to another embodiment of the present invention.

Unless otherwise indicated, the same numbers and symbols in different drawings are usually regarded as corresponding parts. The illustrations are shown to clearly illustrate the relevant associations of the embodiments and not to illustrate the actual dimensions.

In the following, a plurality of embodiments of the present invention will be disclosed graphically. For the sake of clarity, many practical details will be described in the following description. It should be understood, however, that these practical details should not be used to limit the invention. That is, in some embodiments of the present invention, these practical details are unnecessary. In addition, in order to simplify the drawings, some conventional structures and components will be shown in the drawings in a simple and schematic manner.

In this article, the terms first, second, third, etc. are used to describe various elements, components, regions, layers, and / or blocks that are understandable. However, these elements, components, regions, layers and / or blocks should not be limited by these terms. These terms are limited to identifying single elements, components, regions, layers, and / or blocks. Therefore, a first element, component, region, layer, and / or block in the following may also be referred to as a second element, component, region, layer, and / or block without departing from the intention of the present invention.

FIG. 1 is a schematic diagram illustrating a depth image construction system 100 according to various embodiments of the present invention. As shown in FIG. 1, the depth image construction system 100 includes a rotatable mechanism 200, an image capturing device 300, and a computing module 400. The rotatable mechanism 200 is configured to rotate on a plane P1 with a rotation axis RA relative to the circle center C1. In various embodiments, the rotatable mechanism 200 may include a cantilever 220 and a rotation bearing shaft 240 as the rotation axis RA and the circle center C1, respectively, but is not limited thereto. The image capturing device 300 is disposed on the rotatable mechanism 200 and is separated from the circle center C1 by a distance Y. In various embodiments, the distance Y can be used as the rotation radius of the image capture device 300 in subsequent programming. In various embodiments, the distance Y may vary depending on the actual situation. In various embodiments, when the rotatable mechanism 200 is rotated at a plurality of positions A1, A2, A3, A4, etc. of the plane P1 relative to the circle center C1, and the image capturing device 300 is driven, the image capturing device 300 is configured to be at the position At A1, A2, A3, and A4, a plurality of plane images (refer to FIGS. 4A to 4E) of the three-dimensional object (refer to the three-dimensional object 500 in FIG. 3) are acquired, among which a plurality of positions A1, A2, and A3 , A4 are located on the reference circle R1. In various embodiments, the image capturing device 300 may be connected to the computing module 400 in a wireless or wired manner. In various embodiments, captured by the image capture device 300 The acquired planar image may also be stored in other devices and then read by the computing module 400. The computing module 400 may include a first programming 420. In various embodiments, the first programming 420 may generate a depth image according to at least two of the planar images (refer to FIG. 5), which will be described in detail later. For example, the first programming 420 may generate a depth image by using any two of the planar images as the first planar image and the second planar image according to the depth image construction method 700 shown in FIG. 5, but is not limited thereto. .

Because the rotation axis of the depth image construction system 100 rotates on the reference circle R1 with respect to the center C1 at a known distance Y, and is configured to capture a plurality of planes of the three-dimensional object at different known positions A1, A2, A3, and A4, respectively. Therefore, the calculation module 400 can obtain the first program 420 through the calculation of the first programming 420 according to at least two of the foregoing planar images and known conditions as parameters, such as the distance Y, the positions A1, A2, A3, and A4. The magnitude of each depth pixel in a depth image, where the magnitude of the depth pixel can represent the distance of the three-dimensional object from the plane P1. The fixed sampling process and known conditions can reduce the amount of calculation required to execute the first programming 420, so that the computing module 400 can relatively easily execute the first programming 420 to obtain a depth image, and further reduce the construction of the depth image The required time is used to save the computing resources and computing time consumed by the depth image construction system 100 when generating the depth image. In addition, the depth image construction system 100 can achieve the generation of a depth image by using a single image capture device 300, which reduces the construction cost of the depth image construction system 100 and the adjustment time when capturing an image.

FIG. 2 is a schematic diagram illustrating a depth image construction system 100 'according to another embodiment of the present invention. As shown in Figure 2, In the embodiment, the rotatable mechanism 200 'may also include a track 260, and the sliding member 280 is slidably engaged between the track 260 and the image capturing device 300, and is configured to drive the image capturing device 300 to move to the position of the track 260. B1, B2, B3, B4, B5. However, the center of the circle C2 and the axis of rotation RA of the rotatable mechanism 200 'described here are the centers and radii of the reference circle R2 formed by the track 260 on the plane P2, rather than the actual components. In addition, both the positions B1, B2, B3, B4, and B5 can be used as the starting position or the control position described later, which will be described in detail later.

It should be noted that the aforementioned aspect of the depth image construction system 100 or the depth image construction system 100 'is merely an example, and is not intended to limit the present invention. For example, the rotatable mechanism 200 'of the depth image construction system 100' may not be limited to completing the entire circular detour, but may also be a 1/4 circle or a semicircle shape. For example, the track 260 of the depth image construction system 100 'may be other suitable shapes, such as oval or square. It should be understood that those with ordinary knowledge in the field can make appropriate modifications or substitutions without departing from the spirit and scope of the present disclosure, as long as the image capturing device 300 can capture three-dimensional images at different positions on the plane according to actual needs. The plane image of the object, and the rotation radius and rotation angle of each position relative to the center C2 may be known.

FIG. 3 is a simple schematic diagram of a depth image construction system 100 ′ applied to a three-dimensional object 500 in the real world according to various embodiments of the present invention. As shown in FIG. 3, in various embodiments, the normal line N1 of the plane P2 of the center C2 of the reference circle R2 and the three-dimensional object 500 intersect at least a certain point X to ensure that they are located at the positions B1, B2, B3, B4 The plane image captured by the image capturing device 300 of B5 and B5 may include the plane image of the object corresponding to the three-dimensional object 500, such as the plane images 600A ~ 600E included in the plane images 600A to 600E of FIGS. 4A to 4E. Surface images 640A ~ 640E. In various embodiments, the fixed point X is located substantially outside the plane P2 to avoid that the portion where the three-dimensional object 500 coincides with the plane P2 cannot be captured by the image capturing device 300, and may not be able to generate a depth image corresponding to the three-dimensional object 500. That is, in various embodiments, the three-dimensional object 500 needs to be located outside the plane P2, or the plane P2 does not pass through the three-dimensional object 500.

In various embodiments, the normal line of the image capturing device 300 is parallel to the normal line N1 of the plane P2. For example, when the image capture device 300 is located at positions B1, B2, and B3, the normal lines N2, N3, and N4 of the image capture device 300 may be parallel to the normal line N1 of the plane P2, respectively, so that the image capture device 300 The captured planar image of the three-dimensional object 500 can be used to generate depth images in subsequent processes. For example, the depth image construction method 700 shown in FIG. 5 is used to generate the first depth image 800 as shown in FIG. 7 or the second depth image 900 as shown in FIG. 9.

4A to 4E are schematic diagrams of plane images 600A to 600E captured by the three-dimensional object 500 at positions B1 to B5 according to the depth image construction system 100 'according to various embodiments of the present invention. The planar images 600A to 600E in FIGS. 4A to 4E may correspond to the three-dimensional objects 500 captured at positions B1 to B5 in FIG. 3, respectively. In various embodiments, the positions B1 to B5 can be used as the starting position and the reference position, as long as the starting position and the reference position are different. The center angle formed on the reference circle R2 by the respective lines of the starting position and the reference position toward the center C2 of the reference position further allows the first programming 420 to generate a first depth image by using the distance Y and the center angle as parameters. For example, refer to FIG. 3. In various embodiments, the position B1 is used as the starting position, the position B2 is used as the reference position, and the center angle ψ 1 is formed on the reference circle R2. Among others In the embodiment, the position B1 may be used as the starting position, the position B3 may be used as the reference position, and the center angle ψ 2 may be formed on the reference circle R2, but is not limited thereto. As long as the starting position and the reference position do not overlap on the plane P2 and the center angle formed on the reference circle R2 is known, the first programming 420 can be used to generate a first depth image. It should be understood that those with ordinary knowledge in the field may make appropriate modifications or substitutions without departing from the spirit and scope of the present disclosure, according to actual needs.

FIG. 5 is a flowchart of a depth image construction method 700 according to various embodiments of the present invention. 6A to 6C are simple schematic diagrams of different steps in an imaging space in a depth image construction method 700 according to various embodiments of the present invention. As shown in FIG. 5, in various embodiments, the depth image construction method 700 may include steps S710 to S750. Starting from step S710, a first plane image of a three-dimensional object is captured at a starting position of a plane, and the first plane image has a first image baseline. For example, referring to FIGS. 3 and 4B, a planar image 600B (such as FIG. 4B) of the three-dimensional object 500 is captured from the starting position B2 of the plane P2 as the first planar image. Further, the planar image 600B described herein may have a first image baseline 660B.

Step S720 is successively performed, a preset angle is rotated from the starting position relative to the center of the circle on the plane to the control position, and a second plane image of the three-dimensional object is acquired at the control position. The second plane image has a second image baseline. For example, referring to FIGS. 3 and 4C, the preset position Φ is rotated from the starting position B2 to the center C2 on the plane P2 by a predetermined angle Φ to the control position B3, and a plane image 600C of the three-dimensional object 500 is captured at the control position B3 (such as (Figure 4C) as the second plane image. Further, the planar image 600C described herein may have a second image baseline 660C.

Step S730 is successively performed, and the second plane image is rotated according to a preset angle, so that the bottom line of the first image and the bottom line of the second image are parallel to each other. Specifically, refer to FIG. 6A and FIG. 6B. Referring to FIG. 6A, in order to explain the practical significance of step S730, the planar image 600B and the planar image 600C may be further converted to the plane P3 of the imaging space, and the planar image 600B and the planar image 600C are respectively projected on the reference circle R3 of the imaging space. on. The extension line L1 parallel to the edge of the plane image 600B may be at a first angle k between the extension line L1 passing through the center C3 of the reference circle R3 and the extension line L2 of the edge of the parallel plane image 600C may be perpendicular to the center C3 Between the lines VL1, a second angle k + Φ is included. Successively, referring to FIG. 6B, in the plane P3 of the imaging space, the first image floor 660B and the second image floor 660C can be rotated to be parallel to each other by rotating the matrix R. In various embodiments, the second image baseline 660C is rotated to a rotation matrix R parallel to the first image baseline 660B, and the relationship can be: In various embodiments, when the magnitude of θ is the same as the preset angle Φ, the first image baseline 660B and the rotated second image baseline 660C may be parallel to each other.

In other embodiments, step S730 may further include rotating the first plane image according to a correction angle so that the bottom line of the first image and the horizontal line of the three-dimensional object are parallel to each other; and rotating the second plane image according to the correction angle and a preset angle. Make the bottom line of the first image and the bottom line of the second image parallel to each other. For example, as shown in FIG. 6B, in the plane P3 of the imaging space, a rotation matrix R rotation plane image 600B may be first generated according to the correction angle. If the correction angle described here is equal to the first angle k, the first angle k is used as the θ rotation of the rotation matrix R The planar image 600B 'generated by the planar image 600B, the first image baseline 660B of the planar image 600B' in the imaging space, after being rotated, may be substantially parallel to the image baseline 660A of the planar image 600A. If it is referenced together with FIG. 3, it can be seen that the image bottom line 660A is parallel to the horizontal line 520 of the three-dimensional object 500 in space, that is, the first image bottom line of the rotated planar image 600B 'in this embodiment. 660B and the horizontal line 520 of the three-dimensional object 500 are substantially parallel to each other. Successively, another rotation matrix R is used to generate the plane image 600C according to the correction angle and the preset angle Φ, so that the first image baseline 660B and the second image baseline 660C are parallel to each other.

It should be noted that the aforementioned aspect of the rotation matrix R is merely an example, and is not intended to limit the present invention. For example, the first image baseline 660B of the planar image 600B can also be rotated to a direction perpendicular to the horizontal line 520 of the three-dimensional object 500 through the rotation matrix R. It should be understood that those with ordinary knowledge in the field may make appropriate modifications or substitutions without departing from the spirit and scope of the present disclosure, as long as the plane image 600B or the rotated plane image 600B ' The image bottom line 660B and the second image bottom line 660C of the planar image 600C may be parallel to each other.

Step S740 is successively performed, and at least one of the first planar image and the second planar image is translated in a direction perpendicular to the bottom line of the second image, so that the second image bottom line of the second plane image is aligned with the first image bottom line of the first plane image. . For example, referring to FIG. 6C, in various embodiments, on the plane P3 of the imaging space, the plane image 600B 'can be translated along the direction D of the second image baseline 660C after vertical rotation, so that the plane image 660C' The second image baseline 660 of the image is aligned with the first image baseline 660B of the flat image 660B after translation. For example, referring to FIG. 6D, in other embodiments, on the plane P3 of the imaging space, the second plane image 600C 'may also be translated in the direction D of the second image baseline 660C after vertical rotation. The second image baseline 660 of the translated planar image 660C ″ is aligned with the first image baseline 660B of the first planar image 660B ′.

Referring to FIG. 4B and FIG. 4C, in other embodiments, the planar image 600B and the planar image 600C may each include a plurality of planar pixels, such as a planar pixel 620B that does not include an object planar image 640B, and an planar image 640B that includes an object. Plane pixel 620B ', plane pixel 620C that does not include the plane image 640C of the object, and 620C' that includes the plane image 640C of the object, etc. Step S740 may further include calculating a correspondence relationship between the pixel distance of the first planar image and the second planar image in the imaging space and the length in the actual space; and according to the correspondence relationship, calculating and translating the first in a direction perpendicular to the bottom line of the first object image At least one of a planar image and a second planar image. For example, the relative relationship between the plane pixels 620B of the planar image 600B calculated in the imaging space of FIGS. 6A to 6D corresponding to the length in the actual space of FIG. 3 may be used to determine the translation in the imaging space. The pixel distance when the planar image 600B ′ and the planar image 660C ′ are relatively aligned.

Referring to FIG. 6B to FIG. 6D, in other embodiments, the plane image 600B may also include the plane pixel 620B '(e.g., FIG. 4B) of the plane image 640B and the plane image 600C of the object. The relative position of the plane pixels 620C '(as shown in FIG. 4C) of the plane image 640C is determined by the number of pixels required to move the plane image 600B' and the second plane image 600C 'by moving at least one of the plane images 600B' and the second plane image 600C 'in the imaging space. .

FIG. 7 is a schematic diagram of a first depth image 800 according to various embodiments of the present invention. Step S750 is successively performed. After the translation, the distances of the three-dimensional objects relative to the plane of the pixels are respectively calculated according to the first plane image and the second plane image to generate a depth image. For example, referring to FIG. 3, FIG. 6C, and FIG. 7, according to the plane image 600B "and the plane image 600C ', the distances of the captured images in the three-dimensional object 500 and other pixels from the plane P2 are calculated and recorded in The counterpart of the depth pixel 820 in FIG. 7 is to record the distance between the three-dimensional object 500 and the plane P2 to the depth pixel 824, or to record the distance between the image captured in other pixels and the plane P2 to the depth pixel 822, etc. to generate First depth image 800.

For example, in various embodiments, the plane pixels 620B ′ including the plane image 640B of the object, the plane pixels 620C ′ including the plane image 640C of the object, and the depth pixels 824 including the depth image 840 of the object may be relatively positive. The correspondence between the pixels of the planar image 600B and the planar image 600C and the depth pixels 820 is established. The distance of the captured image in the three-dimensional object 500 and other pixels from the plane P2 is further calculated by the method of stereo vision, and is recorded in the depth pixel 820 correspondingly to generate the first depth image 800.

As shown in FIG. 7, the first depth image 800 may include a plurality of depth pixels 820. The magnitude of each depth pixel 820 may represent the distance between the object captured in the corresponding planar image pixel and the plane P2. For example, the depth pixel 820 represents a distance with a value from 0 to 255. A larger value represents a further distance from the plane P2, but is not limited thereto. For example, the first depth image 800 includes the depth pixels of the object depth image 840 corresponding to the three-dimensional object 500 (refer to FIG. 3). 824, whose magnitude represents the distance of the three-dimensional object 500 from the plane P2. In various embodiments, the depth pixels 820 of the first depth image 800 may correspond to the planar pixels of the planar image, for example, the planar pixels 620B, 620B 'of the planar image 600B of FIG. 4B.

Refer back to Figure 6B. In various embodiments, if the plane image 600B ′ is regarded as the first plane image after rotation, and the plane image 600C ′ is regarded as the second plane image after rotation, the circle C3 extends to the first image bottom line 660B. The vertical distance of the line is greater than the vertical distance of the center line C3 to the extension line of the second image bottom line 660C. At this time, the preset angle Φ> 0. In other embodiments, if the plane image 600C ′ is regarded as the first plane image after rotation and the plane image 600B ′ is regarded as the second plane image after rotation, the circle center C3 to the second image bottom line 660B The vertical distance of the extension line of is greater than the vertical distance from the circle center C3 to the extension line of the first image bottom line 660C. At this time, the preset angle Φ <0. It should be understood that those with ordinary knowledge in the field may make appropriate modifications or substitutions without departing from the spirit and scope of the present disclosure, according to actual needs.

FIG. 8 is a simple schematic diagram of different steps in an imaging space in a depth image construction method 700 according to other embodiments of the present invention. FIG. 9 is a schematic diagram of a second depth image 900 according to another embodiment of the present invention. In other embodiments, the depth image construction method 700 further includes step S760. In step S760, the starting position is changed, and steps S710 to S750 are repeatedly performed to obtain a plurality of depth images or at least another depth image, and then average each depth pixel corresponding to the obtained depth image for Updates the depth pixels of a depth image. For example, referring to FIG. 3, FIG. 7 to FIG. 9, the image capturing device 300 moves to positions B1 to B5, and shoots separately Take planar images 600A ~ 600E as shown in Figures 4A to 4E. In various embodiments, the computing module 400 may further include a second programming 440. The second programming 440 may generate a plurality of first depth images 800 according to the foregoing steps S730 to S750 according to the foregoing steps S730 to S750 according to the plurality of planar images, such as the planar images 600A to 600E. Then, the corresponding depth pixels 820 in the plurality of first depth images 800 are averaged to generate a second depth image 900. In other words, the magnitude of the depth pixels 920 in the second depth image 900 is an average of the magnitudes of the corresponding ones in the depth pixels 820 of each first depth image 800.

In summary, the present invention provides a method for constructing a depth image, which captures multiple planar images of a three-dimensional object correspondingly at different positions in a plane, and uses at least two of the plurality of planar images to calculate the distance of the three-dimensional object relative to the plane to Generate depth images. In various embodiments, the present invention may provide a system for applying a method of constructing a depth image, by using a relative relationship between a fixed starting position and a reference position as a known condition, such as a distance between the starting position and the reference position, The rotation angle between the starting position and the reference position relative to the center of the reference circle. Because the change of the sampling process and known conditions is relatively fixed, the system applying the depth image construction method can reduce or reduce the amount of calculation required to generate the depth image, so that the time required for the calculation module to construct the depth image can be further reduced. In order to save the calculation resources and time consumed by the system applying the depth image construction method when generating the depth image. In addition, the system applying the depth image construction method can achieve the generation of the depth image by a single image capture device, reducing the construction cost of the system applying the depth image construction method.

Although the present invention has been disclosed in the above embodiments, it is not intended to limit the present invention. Any person skilled in the art will not depart from the spirit and scope of the present invention. Within the scope, various modifications and retouching can be made, so the protection scope of the present invention shall be determined by the scope of the appended patent application.

Claims (10)

A depth image construction method for generating a first depth image according to a three-dimensional object, wherein the first depth image includes a plurality of depth pixels, and the depth image construction method includes: (a) capturing a starting position in a plane A first plane image of the three-dimensional object, the first plane image having a first image baseline, wherein the normal of the plane is parallel to the normal of the first plane image; (b) relative to the plane from the starting position A center of the circle is rotated by a preset angle to a control position, and a second plane image of the three-dimensional object is captured at the control position. The second plane image has a second image base line, and the normal of the plane is parallel to the The normal of the second plane image; (c) rotating the second plane image according to the preset angle so that the first image baseline and the second image baseline are parallel to each other; (d) in a direction perpendicular to the second image baseline Panning at least one of the first planar image and the second planar image so that the second image baseline of the second planar image is aligned with the first image baseline of the first planar image; and ( e) After the translation, calculate the distance of the three-dimensional object with respect to the plane according to the first plane image and the second plane image, and record the corresponding ones in the depth pixels to generate the first depth image. The method for constructing a depth image according to item 1 of the scope of patent application, wherein step (c) further comprises: (c1) rotating the first plane image according to a correction angle so that the bottom line of the first image and a horizontal line of the three-dimensional object Parallel to each other; and (c2) rotating the second plane image according to the correction angle and the preset angle so that the bottom line of the first image and the bottom line of the second image are parallel to each other. The depth image construction method according to item 1 of the scope of patent application, wherein the normal of the plane passing through the center of the circle intersects the three-dimensional object at least at a certain point. The depth image construction method according to item 3 of the scope of patent application, wherein the fixed point is located outside the plane. The method for constructing a depth image according to item 1 of the scope of the patent application, wherein the first plane image and the second plane image each include a plurality of plane pixels, and step (d) further includes: (d1) calculating the first plane A corresponding relationship between the planar pixels of the image and the planar pixels of the second planar image and the length in space; and (d2) calculating and translating in a direction perpendicular to the bottom line of the first object image according to the corresponding relationship At least one of the first planar image and the second planar image. The method for constructing a depth image as described in item 1 of the patent application scope further comprises: changing the starting position, and repeating steps (a) to (e) to generate another first depth image; and Each of the depth pixels in the first depth image is averaged with the corresponding depth pixel in the first depth image to update the depth pixels of the first depth image to generate a second depth image. A depth image construction system includes: a rotatable mechanism configured to rotate in a plane relative to a circle center; an image capture device disposed in the rotatable mechanism and a distance from the circle center, wherein when the rotatable mechanism When the image capturing device is driven by rotating at a plurality of positions on the plane relative to the center of the circle, the image capturing device is configured to capture a plurality of planar images of a three-dimensional object at the positions, respectively, where the positions can form a A reference circle, and the plane does not pass through the three-dimensional object, the normal of the plane is parallel to the normal of the plane images; and a calculation module including a first programming configured to be based on at least two of the plane images Generate a first depth image. The depth image construction system according to item 7 of the scope of patent application, wherein the normal of the image capturing device is parallel to the normal of the plane. The depth image construction system according to item 7 of the scope of the patent application, wherein the positions corresponding to the plane images include a starting position and a reference position, and the starting position and the reference position are formed along the reference circle. A center angle, and the first programming uses the distance and the center angle to generate the first depth image. The depth image construction system according to item 7 of the scope of the patent application, wherein the depth image includes a plurality of depth pixels, and the calculation module further includes a second programming configured to match the plurality of pairwise generated pairs of the planar images. The first depth image averages the corresponding depth pixels in the first depth images to generate a second depth image.