CN106254855B - A kind of three-dimensional modeling method and system based on zoom ranging - Google Patents

Abstract

The invention discloses a three-dimensional modeling method based on zoom distance measurement, which relates to the field of three-dimensional reconstruction, comprising the following steps: firstly, dividing a target object into several sub-regions; , to obtain the focus image of the sub-region, and collect the focal length or object distance of the sub-region; the zoom evaluation function value is used to measure whether the sub-region is in focus, and the focus image is an image in which the sub-region is in focus; finally, according to the The orthofocus image and the focal length or object distance of each sub-region are used to establish a 3D model. At the same time, the invention also discloses a three-dimensional modeling system based on zoom distance measurement. The invention provides a fast three-dimensional reconstruction method, which avoids the inability to obtain accurate three-dimensional depth information in a local region of the three-dimensional reconstruction by the binocular parallax method. At the same time, the present invention only needs one shooting lens, and the three-dimensional model of the object can be obtained at one position.

Description

A 3D modeling method and system based on zoom ranging

technical field

The invention relates to the field of object three-dimensional reconstruction, in particular to a three-dimensional modeling method and system based on zoom distance measurement.

Background technique

3D reconstruction has a wide range of applications, including volume measurement, stereoscopic display, 3D printing, object model building, etc. The existing 3D reconstruction generally adopts the binocular time-of-flight method, and the object is photographed by two cameras, and the three-dimensional information of the object is obtained and three-dimensionally reconstructed through the slight difference between the two captured images. However, this method has shortcomings: some local areas of the three-dimensional object do not appear in the two images at the same time. When the part of the object appears in the left image and does not appear in the right image, the depth information of the position at this time The processing is often imprecise. To sum up, due to the defects of the technical principle itself in the existing technology, image depth information cannot be obtained well in local areas, which affects the 3D reconstruction of objects, and it is necessary to explore a new method for 3D reconstruction of objects.

Contents of the invention

In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is to provide a three-dimensional modeling method based on zoom ranging, which aims to solve the defect of the prior art due to its technical principle itself, that is, local areas cannot be easily Obtaining image depth information well affects the technical issues of 3D reconstruction of objects.

To achieve the above object, the present invention provides a three-dimensional modeling method based on zoom ranging, including:

Step S1, dividing the target object into several sub-regions;

Step S2, zooming to collect images of each of the sub-regions, obtaining the in-focus image of the sub-region according to the value of the focus evaluation function, and collecting the focal length or object distance of the sub-region; the value of the zoom evaluation function is used to measure the Whether the sub-region is accurately focused, and the in-focus image is an image of the sub-region accurately focused;

Step S3, building a three-dimensional model according to the in-focus image of each of the sub-regions and the focal length or object distance of each of the sub-regions.

In this technical solution, the target is divided into a plurality of sub-regions, each sub-region is focused on, and whether it is focused is judged through image processing, and the in-focus image is acquired and the focal length is recorded. Depth information is obtained according to the focal length or object distance information of each sub-region, and three-dimensional reconstruction is performed. The technical solution provides a fast three-dimensional reconstruction method, which avoids the inability to obtain accurate three-dimensional depth information in a local area of the three-dimensional reconstruction by the binocular parallax method.

Further, the step S2 specifically includes:

Collect the RGB values of M×N pixels corresponding to the sub-region in the ith zoom image of the k-th sub-region J _k ;

Calculate the focus evaluation function value of the i-th zoom image of the sub-region J _k

Obtain the maximum value of the focus evaluation function of the sub-region J _k in one zoom image The focus evaluation function maximum value The corresponding zoom image is the positive focus image of the sub-region; the The I is the total number of zoom images, and I is a natural number;

said

Wherein, the i, k, M, and N are natural numbers, the x satisfies 1≤x≤M, the y satisfies 1≤y≤N, and the R _i (x, y), G _i (x, y ), B _i (x, y) are the RGB values of the pixel respectively; is the average value of R pixels in the sub-region J _k , the is the average value of G pixels in the sub-region J _k , the is the average value of B pixels in the sub-region J _k .

In this technical scheme, through Obtain the focus evaluation function of RGB three colors, and the focus evaluation accuracy is high.

Further, the step S3 further includes: generating sub-region image magnifications according to the object distance or focal length, and correcting the size of each sub-region.

In this technical solution, based on the principle of near-large and far-small, the size of each layer is corrected according to the focal length, that is, the closer the object is to the lens, the smaller the magnification; it is beneficial to improve the accuracy and accuracy of 3D modeling. accuracy.

In view of the defects of the prior art, the second technical problem to be solved by the present invention is to provide a 3D modeling system based on zoom distance measurement, which aims to solve the defects of the prior art due to its technical principle itself, that is, local area The depth information of the image cannot be obtained well, which affects the technical problem of the 3D reconstruction of the object.

To achieve the above object, the present invention provides a three-dimensional modeling method based on zoom ranging, including:

A sub-area division module is used to divide the target object into several sub-areas;

An image acquisition module, configured to zoom in and acquire images of each of the sub-regions;

A zoom drive module, configured to drive the image acquisition module to zoom continuously;

A focal length acquisition module for capturing the focal length or object distance of the sub-area;

The sub-region focus image recognition module is used to obtain the focus image of the sub-region according to the focus evaluation function value; wherein, the zoom evaluation function value is used to measure whether the sub-region is in focus, and the focus image is an in-focus image of said sub-region;

The three-dimensional reconstruction module is used to establish a three-dimensional model according to the in-focus image of each of the sub-regions and the focal length or object distance of each of the sub-regions.

In this technical solution, the target is divided into a plurality of sub-regions, each sub-region is focused on, and whether it is focused is judged through image processing, and the in-focus image is acquired and the focal length is recorded. Obtain depth information according to the focal length or object distance information of each sub-region, and perform 3D reconstruction. The technical solution provides a fast three-dimensional reconstruction method, which avoids the inability to obtain accurate three-dimensional depth information in a local area of the three-dimensional reconstruction by the binocular parallax method.

Further, the sub-region focus image recognition module is configured to:

Collect the RGB values of M×N pixels corresponding to the sub-region in the ith zoom image of the k-th sub-region J _k ;

Calculate the focus evaluation function value of the i-th zoom image of the sub-region J _k

Obtain the maximum value of the focus evaluation function of the sub-region J _k in one zoom image The focus evaluation function maximum value The corresponding zoom image is the positive focus image of the sub-region; the The I is the total number of zoom images, and I is a natural number;

said

Wherein, the i, k, M, and N are natural numbers, the x satisfies 1≤x≤M, the y satisfies 1≤y≤N, and the R _i (x, y), G _i (x, y ), B _i (x, y) are the RGB values of the pixel respectively; is the average value of R pixels in the sub-region J _k , the is the average value of G pixels in the sub-region J _k , the is the average value of B pixels in the sub-region J _k .

In this technical scheme, through Obtain the focus evaluation function of RGB three colors, and the focus evaluation accuracy is high.

Further, the system further includes: a sub-region scaling module; the sub-region scaling module is configured to: generate a sub-region image magnification according to the object distance or focal length, and correct the size of each sub-region .

In this technical solution, based on the principle of near-large and far-small, the size of each layer is corrected according to the focal length, that is, the closer the object is to the lens, the smaller the magnification; it is beneficial to improve the accuracy and accuracy of 3D modeling. accuracy.

The beneficial effects of the present invention are: the present invention divides the target object into a plurality of sub-regions, focuses on each sub-region respectively, judges whether to focus through image processing, acquires a focus image and records the focal length. Depth information is obtained according to the focal length or object distance information of each sub-region, and three-dimensional reconstruction is performed. The technical solution provides a fast three-dimensional reconstruction method, which avoids the inability to obtain accurate three-dimensional depth information in a local area of the three-dimensional reconstruction by the binocular parallax method. At the same time, the present invention only needs one shooting lens, and the three-dimensional model of the object can be obtained at one position.

Description of drawings

Fig. 1 is a schematic flow sheet of a specific embodiment of the present invention;

Fig. 2 is a schematic diagram of the system structure of a specific embodiment of the present invention.

detailed description

Below in conjunction with accompanying drawing and embodiment the present invention will be further described:

As shown in Figure 1, an embodiment of the present invention provides a three-dimensional modeling method based on zoom ranging, including:

Step S1, dividing the target object into several sub-regions;

Step S2, zooming to collect images of each of the sub-regions, obtaining the in-focus image of the sub-region according to the value of the focus evaluation function, and collecting the focal length or object distance of the sub-region; the value of the zoom evaluation function is used to measure the Whether the sub-region is accurately focused, and the in-focus image is an image of the sub-region accurately focused;

Step S3, building a three-dimensional model according to the in-focus image of each of the sub-regions and the focal length or object distance of each of the sub-regions.

This embodiment will be further described below.

In this embodiment, firstly, the target object is divided into several regions; in an optional example, the object image on the shooting device is divided into several regions.

Then, each area is separately focused, and whether the focus is successful is judged by whether the image recognition sharpness or the value of the focus evaluation function calculated by image recognition reaches the maximum.

When the focus is successful, the image of the sub-region captured at this time is the in-focus image. Obviously, the image needs to be segmented to separate the image of the sub-region. Existing commonly used lenses are generally zoomed by a zoom motor, and the zoom motor generally adopts a stepping motor. By collecting the steps of the zoom motor, the zoom focal length or object distance can be obtained. Wherein, the relationship curve between the number of steps of the zoom motor and the focal length may be provided by the manufacturer, or may be obtained through experiments, which is not limited in the present disclosure. Similarly, the relationship curve between the number of steps of the zoom motor and the object distance can also be provided by the manufacturer, or obtained through experiments. It is worth mentioning that a zoom lens can contain two sets of motors, namely the zoom motor and the focus motor. Correspondingly, there can also be a relationship between the focal length and the number of motor steps.

It is worth mentioning that, in addition to using a zoom motor to zoom, there are other zooming methods, such as liquid crystal lens and liquid lens zooming through voltage regulation, and the example of the present disclosure does not limit the zooming method of the lens.

Finally, a three-dimensional model is constructed through the orthofocus images of each area and the corresponding focal length or object distance information.

In an optional example, the depth information of the object sub-region is obtained through the focal length of the sub-region and the focal length-object distance relationship curve to obtain a three-dimensional model.

In another optional example, the focal length is proportional to the object distance, and the focal length information is used to construct a three-dimensional model with a certain proportion of the original object.

In the automatic focusing system, computer programming is used to use some algorithmic rules to judge whether the image clarity has reached the most accurate state, and drive the electric focusing device to focus. This algorithm is called the focusing state evaluation function, referred to as the focusing evaluation function. In the present invention, the focus evaluation function can be used to obtain correctly focused images from multiple zoomed images.

Optionally, the focus evaluation function may be a grayscale gradient function, a frequency domain function, an informatics function, and a statistical function. Algorithms of related evaluation functions are disclosed in the prior art, and will not be repeated here.

Preferably, in the embodiments of the present disclosure, the focus evaluation function used is

Specifically, in this embodiment, the step S2 specifically includes:

Collect the RGB values of M×N pixels corresponding to the sub-region in the ith zoom image of the k-th sub-region J _k ;

Calculate the focus evaluation function value of the i-th zoom image of the sub-region J _k

Obtain the maximum value of the focus evaluation function of the sub-region J _k in one zoom image The focus evaluation function maximum value The corresponding zoom image is the positive focus image of the sub-region; the The I is the total number of zoom images, and I is a natural number;

said

Wherein, the i, k, M, and N are natural numbers, the x satisfies 1≤x≤M, the y satisfies 1≤y≤N, and the R _i (x, y), G _i (x, y ), B _i (x, y) are the RGB values of the pixel respectively; is the average value of R pixels in the sub-region J _k , the is the average value of G pixels in the sub-region J _k , the is the average value of B pixels in the sub-region J _k .

In this embodiment, the depth information of the 3D model is obtained through the focal length, while the size data of the 3D model on the X-Y plane is determined by the size of each sub-region. In fact, according to the near-large-far-small rule, the farther the subject is from the lens, the smaller the image size will be. Therefore, in this embodiment, the size of each sub-region is scaled up according to the object distance or the focal length.

Optionally, the step S3 further includes: generating sub-region image magnifications according to the object distance or focal length, and correcting the size of each sub-region.

In another optional embodiment, the sub-region image magnification β is proportional to the focal length f, and the sub-region image magnification The f ₀ is the focal length corresponding to a magnification of 1. In another optional embodiment, the sub-region image magnification is proportional to the object distance.

As shown in Figure 2, in the second embodiment of the present invention, a 3D modeling system based on zoom ranging is provided, including:

A sub-area division module is used to divide the target object into several sub-areas;

An image acquisition module, configured to zoom in and acquire images of each of the sub-regions;

A zoom drive module, configured to drive the image acquisition module to zoom continuously;

A focal length acquisition module for capturing the focal length or object distance of the sub-area;

The sub-region focus image recognition module is used to obtain the focus image of the sub-region according to the focus evaluation function value; wherein, the zoom evaluation function value is used to measure whether the sub-region is in focus, and the focus image is an in-focus image of said sub-region;

The three-dimensional reconstruction module is used to establish a three-dimensional model according to the in-focus image of each of the sub-regions and the focal length or object distance of each of the sub-regions.

Optionally, the focus evaluation function may be a gray gradient function, a frequency domain function, an informatics function, and a statistical function.

In this embodiment, the sub-area focus image recognition module is configured to:

Collect the RGB values of M×N pixels corresponding to the sub-region in the ith zoom image of the k-th sub-region J _k ;

Calculate the focus evaluation function value of the i-th zoom image of the sub-region J _k

Obtain the maximum value of the focus evaluation function of the sub-region J _k in one zoom image The focus evaluation function maximum value The corresponding zoom image is the positive focus image of the sub-region; the The I is the total number of zoom images, and I is a natural number;

said

Wherein, the i, k, M, and N are natural numbers, the x satisfies 1≤x≤M, the y satisfies 1≤y≤N, and the R _i (x, y), G _i (x, y ), B _i (x, y) are the RGB values of the pixel respectively; is the average value of R pixels in the sub-region J _k , the is the average value of G pixels in the sub-region J _k , the is the average value of B pixels in the sub-region J _k .

In this embodiment, the system further includes: a sub-region scaling module; the sub-region scaling module is configured to: generate a sub-region image magnification according to the object distance or focal length, and correct each of the sub-regions size of.

In this embodiment, the depth information of the 3D model is obtained through the focal length, while the size data of the 3D model on the X-Y plane is determined by the size of each sub-region. In fact, according to the near-large-far-small rule, the farther the subject is from the lens, the smaller the image size will be. Therefore, in this embodiment, the size of each sub-region is scaled up according to the object distance or the focal length.

Optionally, the step S3 further includes: generating sub-region image magnifications according to the object distance or focal length, and correcting the size of each sub-region.

In another optional embodiment, the sub-region image magnification β is proportional to the focal length f, and the sub-region image magnification The f ₀ is the focal length corresponding to a magnification of 1. In another optional embodiment, the sub-region image magnification is proportional to the object distance.

In this embodiment, the shooting device is aimed at the target object, the image acquisition module obtains the image of the target object, and the sub-region division module divides the influence of the target object into multiple sub-regions, and each sub-region is sequentially selected for zooming to collect images; After collecting a zoom image, the focal length of the image is collected and recorded by the focal length acquisition module at the same time. Whenever all the zoom images of a sub-region are collected, the sub-region positive-focus image recognition module selects the zoom image with the highest value of the focus evaluation function as the positive-focus image, and intercepts the region image corresponding to the sub-region on the image. Then, the sub-area scaling module scales up the image according to the focal length of the front-focus image to obtain the X-Y plane size. Perform 3D reconstruction of the target object through focal length, object distance or depth information and X-Y plane size.

The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make many modifications and changes according to the concept of the present invention without creative efforts. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning or limited experiments on the basis of the prior art shall be within the scope of protection defined by the claims.

Claims (3)

1. a kind of three-dimensional modeling method based on zoom ranging, it is characterised in that methods described includes：

Step S1, target object is divided into some subregions；

Step S2, the image of each subregion is gathered in same position zoom, according to focusing evaluation function value, described in acquisition The positive burnt image of subregion, and gather the focal length or object distance of the subregion；The zoom evaluation function value is described for weighing Whether subregion accurately focuses, and the just burnt image is the image that the subregion is accurately focused；

Step S3, according to the positive burnt image of each subregion and the focal length or object distance of each subregion, establish three-dimensional Model；

The step S2, is specifically included：

Gather k-th of subregion J_kI-th zoom image in the M × N number of pixel corresponding with the subregion rgb value；

Calculate the subregion J_kI-th zoom image focusing evaluation function value

Obtain the subregion J_kFocusing evaluation function maximum in the I zoom imagesThe focusing evaluation Function maximaCorresponding zoom image is the positive burnt image of the subregion；It is describedIt is described I is zoom image sum, and I is natural number；

It is described

Wherein, described i, k, M, N are natural number, and the x meets that 1≤x≤M, the y meet 1≤y≤N, the R_i(x,y)、G_i (x,y)、B_i(x, y) is respectively the rgb value of pixel；It is describedFor subregion J_kInterior R pixel averages, it is describedFor subregion J_kIt is interior G pixel averages, it is describedFor subregion J_kInterior B pixel averages.

A kind of 2. three-dimensional modeling method based on zoom ranging as claimed in claim 1, it is characterised in that the step S3, Also include：According to the object distance or focal length, sub-district area image enlargement ratio is generated, corrects the size of each subregion.

A kind of 3. 3 d modeling system based on zoom ranging, it is characterised in that including：

Sub-zone dividing module, for target object to be divided into some subregions；

Image capture module, for gathering the image of each subregion in same position zoom；

Zoom drive module, for driving described image acquisition module continuous vari-focus；

Shooting focal length acquisition module, for gathering the focal length or object distance of the subregion；

The just burnt picture recognition module of subregion, for obtaining the positive burnt image of the subregion according to focusing evaluation function value；Its In, the zoom evaluation function value is used to weigh whether the subregion accurately focuses, and the just burnt image is the subregion The image accurately focused；

Three-dimensional reconstruction module, for the positive burnt image and the focal length or thing of each subregion according to each subregion Away from, and establish threedimensional model；

The just burnt picture recognition module of the subregion, is configured as：

Gather k-th of subregion J_kI-th zoom image in the M × N number of pixel corresponding with the subregion rgb value；

Calculate the subregion J_kI-th zoom image focusing evaluation function value

Obtain the subregion J_kFocusing evaluation function maximum in the I zoom imagesThe focusing evaluation Function maximaCorresponding zoom image is the positive burnt image of the subregion；It is describedIt is described I is zoom image sum, and I is natural number；

It is described

Wherein, described i, k, M, N are natural number, and the x meets that 1≤x≤M, the y meet 1≤y≤N, the R_i(x,y)、G_i (x,y)、B_i(x, y) is respectively the rgb value of pixel；It is describedFor subregion J_kInterior R pixel averages, it is describedFor subregion J_kIt is interior G pixel averages, it is describedFor subregion J_kInterior B pixel averages.