CN100573040C - The scaling method of object surface three-dimensional contour structure light measurement system - Google Patents

Abstract

A kind of scaling method of object surface three-dimensional contour structure light measurement system belongs to method for three-dimensional measurement and Instrument technology field.It is made up of following steps: (1) black and white camera calibration; (2) projector calibrating; (3) many plane references method obtains the required parameter of total system; It is characterized in that: (4) described target plate is colored target plate, and the prospect on the target plate and background colour are two kinds of two kinds of colors that the black and white camera had same reflecting properties; (5) when (1) goes on foot the black and white camera calibration, throw a kind of colorama to the target plate, this colorama is identical with one of background colour with target plate prospect.The inventive method has higher stability, reliability than traditional use black and white target plate, has improved stated accuracy, accurately important effect is arranged aspect the measurement at object surface three-dimensional contour structure light.

Description

Calibration method of structured light measurement system for three-dimensional contour of object surface

Technical field

A calibration method for a three-dimensional profile structured light measurement system on an object surface relates to the field of computer vision measurement, belongs to the technical field of three-dimensional measurement methods and instruments, and in particular relates to the calibration occasion for a structured light measurement system.

Background technique

The three-dimensional measurement technology of objects should be very common in the fields of product design and manufacturing, quality inspection and control, robot vision, etc. Structured light measurement technology is a very important and promising branch of 3D measurement technology. Accurate calibration of the measurement system is the prerequisite and key condition for the system to be accurately reconstructed. The calibration of the measurement system is to obtain the geometric and optical characteristics inside the camera and projector, that is, the internal parameters, and the positional relationship between the coordinate system of the two instruments relative to the spatial coordinate system, that is, the external parameters.

The calibration of the structured light measurement system includes two parts, one is the calibration of the camera, and the other is the calibration of the projector. The camera calibration is currently based on Tsai’s Radial Constraint Consistency (RAC) [R.Y.Tsai, "A Versatile Camera Calibration Technique for High-Accuracy 3D Machine Vision Metrology Using Off-the-shelf TVCameras and Lenses," IEEE Journal of Robotics Automation. Vol.RA-3.No.4, August 1987.] is the most commonly used method. This method first solves the linear parameters of the camera model by radial alignment constraints, then introduces nonlinear parameters, and uses an iterative method to find the overall parameters. There is also some research on the calibration of projectors. The most widely used method is to calibrate the projector as a reverse model of the camera. The calibration method is similar to that of the camera. However, there are many restrictions on calibrating a structured light measurement system integrated with a camera and a projector. The circular mark point is used as the target pattern, and the specific calibration method is as follows: firstly, the camera is calibrated, and the camera collects the target point image in the known coordinate space, through the noise filtering of the image, threshold segmentation, contour extraction, ellipse recognition, ellipse fitting and The center is extracted to obtain the two-dimensional image coordinates of the three-dimensional target points, and the corresponding three-dimensional and two-dimensional point pairs are matched as input data, and the calibration method of Tsai is used to calibrate the camera; then, the projector is calibrated: because the projector does not have image acquisition Therefore, the calibration of the projector should use auxiliary structured light or other encoding modes to indirectly obtain the 2D and 3D target point pairs for projector calibration. Method 1: Ensure that the position of the target plate used for camera calibration is known, the projector projects a circular or square pattern with known coordinates onto the target plate for camera calibration, the camera collects the projected pattern, detects the coordinates of the center of the circle, and uses the calibration The camera parameters and the Z-direction coordinates of the target board are used to reconstruct the spatial three-dimensional target point coordinates of the logo pattern projected by the projector, and the corresponding two-dimensional and three-dimensional point pairs are used as calibration input data; the second method is to ensure the target board used for camera calibration The position is known, the projector projects the coded stripes of structured light onto the target plate, the camera collects the coded stripes, and uses (Gray+phase) Gray code + phase shift technology to decode the coded value of the two-dimensional pixel coordinates of each target point. Match the coding value of each target point into the projector, and the pixel coordinates in the projector are the two-dimensional pixel coordinates of the three-dimensional target point, realizing the function of "grabbing" the image of the projector. Match the corresponding 2D and 3D point pairs as input data to calibrate the projector. From the above method, it is found that in the calibration of the projector, the position of the target is guaranteed to be known, the projector projects the image, and the camera collects the image, that is, the target point on the target is used for the calibration of the camera and the projector at the same time, and is used for the camera and the projection. The calibration pattern of the projector is different, so when the coding mode is projected on the traditional black and white target pattern for projector calibration, the collected target image will be affected by the target pattern used to calibrate the camera, and a good target pattern cannot be obtained, seriously affect the calibration accuracy.

Contents of the invention

Aiming at the deficiencies of the existing methods, a calibration method for the three-dimensional contour structured light measurement system of the object surface using a colored target pattern is proposed, which solves the above problems well according to the optical characteristics, and ensures that the collected target pattern is in contrast The high black and white pattern makes the table setting more stable and reliable.

A calibration method for a three-dimensional profile structured light measurement system on an object surface, comprising the following steps:

(1) Black-and-white camera calibration: the camera collects clear target images for camera calibration, uses image processing methods to extract the two-dimensional pixel coordinates of the target points, and combines the known three-dimensional coordinates of the target points with the corresponding two-dimensional pixels The coordinates are stored as the calibration input data of the target plate at the spatial position;

(2) Projector calibration: keep the target board at this space position, the target board is illuminated by white light, the projector projects a black and white coded grating onto the target board, the camera collects the target image with grating stripes, and uses decoding technology to obtain each The coded value of a pixel point, according to the coded value corresponds to the pixel coordinate with the same coded value in the projector, obtains the pixel coordinate of the point in the projector. Store the three-dimensional coordinates of the point and the corresponding one-dimensional pixel coordinates in the projector as the calibration input data of the target plate at the spatial position;

(3) Move the target plate for a certain distance precisely in the direction perpendicular to the target plate and repeat steps (1) and (2), move at least three times, use the multiple sets of data obtained as calibration input data, and use the multi-plane calibration method Carry out system calibration to obtain the final calibration parameters of cameras and projectors;

It is characterized by:

(4), the target plate is a color target plate, and the foreground and background colors on the target plate are two colors that have the same reflective properties for black and white cameras. When the projector is calibrated: the color target plate is illuminated by white light, the projector projects a grating pattern onto the target plate, and the black and white CCD camera can collect a grating-coded fringe image with good grayscale information that is not affected by the pattern of the calibration plate itself;

(5) When calibrating the black-and-white camera in step (1), project a colored light onto the target board, which is the same as one of the foreground and background colors of the target board. When calibrating a black-and-white camera: Because a color target pattern only reflects light of the same color as it, the color calibration pattern captured in a black-and-white CCD camera is a black-and-white image with good black-and-white contrast. The target pattern with the same color as the projected light appears white in the black-and-white image, and the target pattern with a different color from the projected light appears black in the black-and-white image. The projection of colored light described in this step can be accomplished by using a projector, so as to avoid the complexity of the device.

The present invention uses two colors that have the same emissivity to the black-and-white camera as the foreground color and the background color of the target plate respectively. For the traditional black and white target board, when the projector is calibrated, the black and white target image interferes with the acquisition of grayscale information of the coded raster image, which affects the solution for phase decoding. Using colored targets avoids this problem.

On the basis of using a color target plate, using the optical characteristics to project a color light that is the same as the foreground or background color of the target plate to the color target plate, the black and white camera can obtain the same quality effect as the traditional black and white target plate. Strong contrast black and white image.

Compared with the traditional black-and-white target plate method, the method of the invention has higher stability and reliability, and improves the calibration precision.

Description of drawings

FIG. 1 is a schematic diagram of a circular color target used for system calibration in the present invention. The circular target object, that is, the foreground color is red, and the background color of the target plate is blue.

Fig. 2 is an image of the target plate collected by a black and white camera under the irradiation of red light for the color target plate of the present invention.

Figure 3 is the grating stripe target image presented by the traditional target plate in the camera.

Fig. 4 is a target image of grating stripes projected onto the target plate by a projector captured by a camera under white light irradiation of the color target plate of the present invention.

Fig. 5 is a principle diagram of establishing three-dimensional coordinates of a target plate of the present invention.

specific implementation plan

The method of the present invention will be described in further detail below in conjunction with the accompanying drawings. The invention uses a simple color plane target for the first time to calibrate all the parameters of the camera and the projector of the structured light vision measurement system.

Target board fabrication. Fig. 1 is the colored target of the present invention, and circular target thing is that foreground color is red, and target plate background color is that background color is blue. The target is to print colored target dots on a diffuse reflective cellophane and paste them on a flat plate with a flatness less than 0.01. The target dot array is M×N, the rows and columns are strictly vertical, and the center distance of the circular holes is equal and known (Δx , Δy), the difference from the traditional target is that the target of the present invention is in color, and the black-and-white gray-scale target image with the same effect as the traditional black-and-white target can be obtained under a certain light. The straight line where the horizontal center of the target image is located is the X-axis, the straight line where the vertical center is located is the Y-axis, and the direction perpendicular to the target plate is the Z-axis to establish a spatial coordinate system that conforms to the right-hand rule. The radius of the large circle near the center in the figure is 1.5 or 2 times that of the small circle, ensuring that the two can be clearly distinguished for easy detection and sorting.

Calibration pre-processing work. Turn on the camera and projector, adjust the target plate in the common field of view of the camera and projector and the best depth of field, the projector projects a red light source onto the target plate, adjust the camera lens so that the camera can see a target pattern with good contrast , and the target image is clear enough. The projector projects the grating stripes, and by adjusting the camera iris, the camera sees a target pattern with the grating stripes that has good grayscale information.

Camera calibration input data acquisition. The projector projects a red light source onto the target board, and the red color is the same as the foreground color in the target board. Due to the characteristics of optics, because a color target pattern only reflects light with the same color as it, then the color calibration pattern captured in the black and white CCD camera For black and white. The camera collects a clear target image with good black and white contrast, image noise filtering, threshold segmentation, contour extraction, ellipse recognition, ellipse fitting and center extraction, and obtains a sub-two-dimensional image of the target point on the target plate with known three-dimensional coordinates. Pixel center coordinates. The three-dimensional and two-dimensional coordinate pairs of the markers on the target plate are stored as calibration input data. The collected black and white calibration patterns are shown in Figure 2.

Projector calibration input data acquisition. The projector does not have the function of collecting images, therefore, it is necessary to obtain the image point of the calibration point on the projector through an indirect method. The projector projects the coded grating onto the target, and uses the method of coded light and phase shift to obtain the coded value of each point on the captured image, and finds the image point of the calibration point in the projector according to the coded value. In order to ensure that the same pixel coordinate point captured twice by camera calibration and projector calibration corresponds to the same 3D target point, and to ensure that the position of the target plate remains unchanged in this space, the projector projects the encoded grating stripe light onto the target plate, and the camera collects the encoded grating The pattern is shown in Figure 4. The raster image collected by the camera is not affected by the underlying red and blue target plate used for camera calibration, and good grayscale information of the grating stripes is obtained, avoiding the traditional black and white bottom layer used for camera calibration. Influenced by the target pattern, the raster image is coded and decoded through image processing to obtain the code value of each pixel, because when the projector projects the raster image, there is the same code value on the projector screen, and the camera and the projection The pixel coordinates with the same code value in the instrument are matched to realize the matching of the image of the same three-dimensional coordinate point in the camera and the projector, so that the projector has the function of "grabbing" the image, thus obtaining the three-dimensional target point An image point within a projector. In this way, the acquisition function of a location target image is realized.

Get calibration input data for multiple locations. Move the target plate (z distance) along the z-axis direction, that is, the direction perpendicular to the target plate, repeat the above steps, and collect the image points of the calibration input data of the camera and projector at a new position. In order to obtain more accurate calibration parameters , the position of the target plate in the space should try to cover a larger space area as shown in Figure 5.

Calibrate to obtain system parameters. The corresponding three-dimensional target points and corresponding image points in the space multi-plane are used as input data, and Tsai's RAC multi-plane calibration method is used for system calibration to obtain the parameters required for system measurement. In this way, the precise calibration of the three-dimensional profile structured light measurement system on the object surface is realized.

Claims (3)

1, a kind of scaling method of object surface three-dimensional contour structure light measurement system, form by following steps:

(1), black and white camera calibration: camera collects and is used for camera calibration target image clearly, utilize image process method, extract the two-dimensional pixel coordinate of target point, known target point three-dimensional coordinate and corresponding two-dimensional pixel coordinate are stored in the demarcation input data of this locus as the target plate;

(2), projector calibrating: keep the target plate motionless in this locus, the target plate is under the irradiation of white light, projector projects black and white encode grating striped is to the target plate, the camera collection has the target image of grating fringe, utilize decoding technique to obtain the encoded radio of each pixel, correspond to the pixel coordinate that has the same-code value in the projector according to encoded radio, obtain this pixel coordinate in projector, the three-dimensional coordinate of this point and in projector corresponding one dimension pixel coordinate store in the demarcation input data of this locus as the target plate;

(3), the target plate is accurately moved a certain distance along the direction perpendicular to the target plate, repetition (1), (2) step, at least move three times, the multi-group data that obtains is imported data as demarcating, use many plane references method to obtain the parameter of camera and the final required demarcation of projector;

It is characterized in that:

(4), described target plate is colored target plate, and the prospect on the target plate and background colour are two kinds of two kinds of colors that the black and white camera had same reflecting properties;

(5), in (1) step during the black and white camera calibration, throw a kind of colorama to the target plate, this colorama is identical with one of background colour with target plate prospect.

2, the scaling method of object surface three-dimensional contour structure light measurement system according to claim 1 is characterized in that: described (3) step many plane references method be Tsai based on radial constraint unanimity (RAC) method.

3, the scaling method of object surface three-dimensional contour structure light measurement system according to claim 1 and 2 is characterized in that: the described projection of (5) step is colored to utilize projector to finish.

Images