High-precision laser line optical center extraction method and system based on FPGA

Abstract

The application provides a high-precision laser line optical center extraction method and system based on FPGA. After obtaining the laser line image to be extracted, the average width of the laser line in the laser line image to be extracted is calculated, the low-pass filtering parameter is calculated according to the average width, the low-pass filtering processing is performed on the laser line image to be extracted according to the low-pass filtering parameter, and the filtered laser line image is obtained. The preliminary optical center position is located in the filtered laser line image, and the image data in the preset extraction range around the preliminary optical center position is extracted. The image data is fitted to obtain the laser line optical center. The method can pre-process the laser line image to be extracted, filter out image noise, thereby ensuring the robustness and applicability of the subsequent optical center extraction algorithm, and improving the precision and speed of the line optical center extraction.

Description

Technical Field

[0001] This application relates to the field of 3D vision technology, and in particular to a high-precision laser line optical center extraction method and system based on FPGA. Background Technology

[0002] Line laser 3D vision measurement systems are widely used in the field of 3D imaging and measurement due to their simple principle and ease of implementation. Two important parameters of a line laser 3D vision measurement system are accuracy and speed, which determine the system's performance.

[0003] In online laser 3D vision systems, the measurement accuracy is primarily determined by the camera resolution, the system's structural relationships, and the accuracy of line laser center extraction. While camera resolution and structural relationships can be determined through selection and design and remain largely unchanged afterward, the accuracy of line laser center extraction directly determines the measurement accuracy of the line laser 3D vision system.

[0004] However, the measurement accuracy of line laser 3D vision systems is affected by external factors. The main factors influencing the accuracy of line laser center extraction include the width and uniformity of the laser line itself, the properties of the measured object's surface, the influence of external background light sources, and noise introduced during image acquisition. Therefore, to improve the accuracy of line laser center extraction, it is necessary to improve the uniformity and stability of the laser line itself, reduce the influence of background light sources, and filter noise introduced by object surface reflection and imaging units during the acquisition process. Thresholding methods, curve fitting for extreme values, and centroid methods can be used to extract the line laser center. These methods can improve the optical center extraction accuracy to some extent in specific scenarios, but they still suffer from low robustness and limited applicability, resulting in relatively low optical center extraction accuracy. Furthermore, in line laser 3D vision systems, the measurement speed is mainly determined by the camera chip's acquisition speed and the optical center extraction algorithm's speed. With the widespread adoption of high-speed camera chips, the flip-up speed of camera chips is increasing, which in turn requires increasingly faster processing speeds for the optical center extraction algorithm. Solutions using only a central processing unit require a long computation time, which cannot meet the requirements of fast, real-time online measurement. Summary of the Invention

[0005] This application provides a high-precision laser line center extraction method and system based on FPGA to solve the problems of low precision and low speed in line center extraction.

[0006] In a first aspect, this application provides a high-precision laser line center extraction method based on FPGA, comprising:

[0007] Acquire the image of the laser line to be extracted;

[0008] Calculate the average width of the laser lines in the laser line image to be extracted;

[0009] Calculate the low-pass filter parameters based on the average width;

[0010] The laser line image to be extracted is subjected to low-pass filtering processing according to the low-pass filtering parameters to obtain a filtered laser line image;

[0011] Locate the initial optical center position in the filtered laser line image, and extract image data within a preset extraction range around the initial optical center position;

[0012] The image data is fitted to obtain the laser line center.

[0013] In an optional implementation, before the step of calculating the average width of the laser lines in the laser line image to be extracted, the method further includes:

[0014] The laser line image to be extracted is binarized to generate a binarized image;

[0015] Calculate the average width of the laser lines in the binarized image.

[0016] In an optional implementation, the step of calculating the average width of the laser lines in the binarized image includes:

[0017] Traverse the pixels in the binarized image in the column direction;

[0018] The number of consecutive pixels with a pixel value greater than zero in each column of the binarized image is calculated to obtain the laser line column width, which is the maximum number of consecutive pixels in the column.

[0019] The laser line array widths are sorted in ascending or descending order according to their width values ​​to generate a width array.

[0020] The average width is calculated based on the width sequence, where the average width is the median of the width sequence.

[0021] In an optional implementation, the method further includes:

[0022] Calculate the cutoff frequency of the low-pass filter based on the average width;

[0023] Image noise with a frequency greater than the cutoff frequency is filtered out from the laser line image to be extracted in order to obtain a filtered laser line image.

[0024] In an optional implementation, the step of performing low-pass filtering processing on the laser line image to be extracted according to the low-pass filtering parameters includes:

[0025] Perform a Fast Fourier Transform on the laser line pattern to be extracted to generate a spectrum;

[0026] Move the zero-frequency domain portion of the spectrum to the center of the spectrum;

[0027] Set the Gaussian low-pass filter according to the low-pass filter parameters described above;

[0028] The spectrum is subjected to low-pass filtering based on a Gaussian low-pass filter, and the zero-frequency domain portion is restored to its initial position after low-pass filtering.

[0029] Perform an inverse fast Fourier transform on the spectrum to generate a filtered laser line image.

[0030] In an optional implementation, the step of locating the preliminary optical center position in the filtered laser line image includes:

[0031] The preliminary optical center position is calculated based on the following formula:

[0032]

[0033] Where h represents the initial optical center position, hi represents the row number corresponding to the pixel, gi represents the gray value of the pixel, and k is the total number of pixels in the column direction.

[0034] In an optional implementation, the step of extracting image data within a preset range around the initial optical center position further includes:

[0035] The extraction range is calculated based on the average width, and the extraction range is half of the average width.

[0036] Image data within the extraction range above and below the initial optical center position is extracted from the filtered laser line image.

[0037] In an optional implementation, the step of fitting the image data includes:

[0038] Gaussian curve fitting is performed on the image data to obtain a Gaussian curve;

[0039] The optical center of the laser line is extracted based on the Gaussian curve, where the optical center of the laser line is the center of the Gaussian curve.

[0040] Secondly, this application provides a high-precision laser line optical center extraction system based on FPGA, including an image acquisition unit, an image preprocessing unit, and a laser line optical center extraction unit, wherein:

[0041] The image acquisition unit is used to acquire images of the laser line to be extracted;

[0042] An image preprocessing unit is used to calculate the average width of the laser line in the laser line image to be extracted; calculate low-pass filtering parameters based on the average width; and perform low-pass filtering processing on the laser line image to be extracted according to the low-pass filtering parameters to obtain a filtered laser line image.

[0043] The laser line optical center extraction unit is used to locate the initial optical center position in the filtered laser line image, and to extract image data within a preset extraction range around the initial optical center position; and to fit the image data to obtain the laser line optical center.

[0044] In an optional implementation, the image preprocessing unit is further configured to: obtain a pixel threshold;

[0045] Traverse the pixels in the laser line image to be extracted in the column direction;

[0046] The number of consecutive pixels in each column of the laser line image to be extracted that have a pixel value greater than the pixel threshold is calculated to obtain the laser line column width, which is the maximum number of consecutive pixels in the column.

[0047] The laser line array widths are sorted in ascending or descending order according to their width values ​​to generate a width array.

[0048] The average width is calculated based on the width sequence, where the average width is the median of the width sequence.

[0049] As can be seen from the above technical solutions, this application provides a high-precision laser line optical center extraction method and system based on FPGA. The method, after acquiring an image of the laser line to be extracted, calculates the average width of the laser line in the image, calculates low-pass filtering parameters based on the average width, and performs low-pass filtering processing on the laser line image to be extracted according to the low-pass filtering parameters to obtain a filtered laser line image. The method then locates the initial optical center position in the filtered laser line image and extracts image data within a preset extraction range around the initial optical center position. The image data is then fitted to obtain the laser line optical center. This method can preprocess the laser line image to be extracted to filter out image noise, thereby ensuring the robustness and applicability of the subsequent optical center extraction algorithm, while improving the accuracy and speed of line optical center extraction. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 A schematic flowchart of the laser line center extraction method provided in the embodiments of this application;

[0052] Figure 2 A schematic diagram of the laser line image to be extracted provided in an embodiment of this application;

[0053] Figure 3 A schematic diagram of a filtered laser line image provided in an embodiment of this application;

[0054] Figure 4 A schematic diagram of the process for extracting the optical center of a laser line provided in an embodiment of this application;

[0055] Figure 5 This is a structural block diagram of a laser line center extraction method apparatus provided in an embodiment of this application. Detailed Implementation

[0056] The embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described below do not represent all embodiments consistent with this application. They are merely examples of systems and methods consistent with some aspects of this application as detailed in the claims.

[0057] To improve the extraction accuracy of line laser centers, thresholding, curve fitting for extrema, and centroid methods can be used. While these methods can improve the extraction accuracy in specific scenarios, they still suffer from low robustness and limited applicability, resulting in relatively low extraction accuracy. Furthermore, in online laser 3D vision systems, the measurement speed is primarily determined by the camera chip's acquisition speed and the speed of the optical center extraction algorithm. With the increasing prevalence of high-speed camera chips, their flip-flop speeds are accelerating, demanding ever-increasing processing speeds from the optical center extraction algorithm. Solutions relying solely on a central processing unit (CPU) require excessive computation time, failing to meet the requirements of rapid, real-time online measurement.

[0058] To improve the accuracy and speed of line optical center extraction, some embodiments of this application provide a high-precision laser line optical center extraction method based on FPGA. This method can preprocess the image of the laser line to be extracted, filtering out image noise, thereby ensuring the robustness and applicability of the subsequent optical center extraction algorithm, while improving the accuracy and speed of line optical center extraction. Figure 1 As shown, Figure 1 This is a schematic flowchart of a laser line center extraction method provided in an embodiment of this application. The laser line center extraction method includes the following steps:

[0059] S100: Acquire the image of the laser line to be extracted.

[0060] The laser line image to be extracted refers to the image for which the optical center of the laser line needs to be extracted. This application does not specifically limit the implementation method for acquiring the laser line image to be extracted; those skilled in the art can adjust it according to the actual situation. For example, the laser line image to be extracted can be received from other devices, or the laser line image to be extracted can be acquired based on a graphics sensor.

[0061] S200: Calculate the average width of the laser lines in the laser line image to be extracted.

[0062] To improve the accuracy and speed of laser line center extraction, after acquiring the image of the laser line to be extracted, binarization processing can be performed on the image to generate a binarized image. Then, the average width of the laser line in the binarized image can be calculated. By performing binarization processing on the image and then performing subsequent operations on the binarized image, the amount of image processing data is reduced, and the processing speed is improved.

[0063] It should be noted that the embodiments of this application do not specifically limit the implementation of binarization processing. For example, binarization processing methods such as the averaging method, the bimodal method, and the Otsu method can be used. Taking the averaging method as an example, the pixel values ​​of each pixel in the laser line image can be added together, and then the average pixel value of the entire image can be calculated. Then, the entire image is traversed, and the current pixel value is compared with the average pixel value to obtain the binarized value of each pixel.

[0064] In some embodiments, when calculating the average width of the laser line in the image to be extracted, the pixels in the binarized image can be traversed along the column direction. The number of consecutive pixels with a value greater than zero in each column of the binarized image can be calculated to obtain the laser line column width. The laser line column width is the maximum number of consecutive pixels in the column. That is, a pixel threshold is preset, and when calculating the laser line column width, this pixel threshold is called, and the number of consecutive pixels with a value greater than the pixel threshold in each column of the image to be extracted is calculated. The maximum number of consecutive pixels in this column is then selected as the width of this column, i.e., the laser line column width.

[0065] The laser line column widths are then sorted in ascending or descending order to generate a width column. The average width is calculated based on this width column, where the average width is the median of the width column. In other words, after obtaining the laser line column widths for all columns, all laser line column widths are sorted, and the median value is taken as the average width of the laser lines in the image to be extracted.

[0066] S300: Calculates low-pass filter parameters based on average width.

[0067] After obtaining the average width, the low-pass filter parameters can be calculated based on the average width. The low-pass filter parameters can be the cutoff frequency for performing low-pass filtering on the laser line image to be extracted. For example, the low-pass filter parameters can be calculated using the following formula:

[0068] D0=a1k+a2

[0069] Where D0 represents the cutoff frequency of the low-pass filter, k represents the average width, and a1 and a2 are preset parameters.

[0070] S400: Perform low-pass filtering on the laser line image to be extracted according to the low-pass filtering parameters to obtain a filtered laser line image.

[0071] After calculating the low-pass filter parameters based on the average width, the low-pass filter image to be extracted can be processed by low-pass filtering according to the low-pass filter parameters. For example, if the low-pass filter parameter is the cutoff frequency of the low-pass filter, after calculating the cutoff frequency of the low-pass filter based on the average width, the low-pass filter image to be extracted can be processed by low-pass filtering to filter out image noise with a frequency greater than the cutoff frequency in the image to be extracted, so as to obtain the filtered laser line image.

[0072] In some embodiments, low-pass filtering can be performed on the laser line image to be extracted. First, a fast Fourier transform is performed on the laser line image to be extracted to generate a spectrum. The zero-frequency domain portion of the spectrum is moved to the center of the spectrum, and the low-pass filter is set according to the low-pass filter parameters, where the zero-frequency domain portion is the signal with a frequency of zero in the spectrum.

[0073] Low-pass filtering is applied to the spectrum. The spectrum is input into a low-pass filter, which removes image noise with frequencies higher than the cutoff frequency from the laser line image to be extracted, resulting in the noise-free output spectrum. Simultaneously, the zero-frequency domain portion is restored to its initial position after low-pass filtering. The initial position refers to the position of the zero-frequency domain portion before it was moved. Finally, an inverse Fast Fourier Transform (IFFT) is performed on the low-pass filtered spectrum to generate the filtered laser line image.

[0074] It should be noted that the embodiments of this application do not specifically limit the type of low-pass filter, and those skilled in the art can adjust it according to the actual situation. For example, it can be a Gaussian low-pass filter, a median low-pass filter, a bilateral low-pass filter, a Wiener low-pass filter, a Kalman low-pass filter, etc. Taking a Gaussian low-pass filter as an example, when performing low-pass filtering processing on the laser line image to be extracted, the Gaussian low-pass filter can be set according to the low-pass filtering parameters, and low-pass filtering can be performed on the spectrum based on the Gaussian low-pass filter. For example, the Gaussian low-pass filter is as follows:

[0075]

[0076] Where H(u, v) is the Gaussian low-pass filter function, D0 represents the cutoff frequency, and D(u, v) represents the distance from point (u, v) to the center point. The spectrum corresponding to the laser line pattern to be extracted is input into this Gaussian low-pass filter to obtain the spectrum of the image after filtering out image noise from the Gaussian low-pass filter output.

[0077] like Figure 2 The diagram illustrates an image of a laser line to be extracted, provided in an embodiment of this application. This image is input into the aforementioned Gaussian low-pass filter. The Gaussian low-pass filter performs low-pass filtering on the image to be extracted, thereby obtaining the filtered laser line image output by the Gaussian low-pass filter. Figure 3 The image shown is a schematic diagram of a filtered laser line image provided in an embodiment of this application. According to... Figure 2 , Figure 3 It can be seen that the image after low-pass filtering is smoother, thus ensuring the robustness and applicability of the subsequent optical center extraction algorithm.

[0078] S500: Locate the initial optical center position in the filtered laser line image.

[0079] After obtaining the filtered laser line image, the preliminary optical center position can be located in the filtered laser line image based on the centroid method. For example, the preliminary optical center position can be calculated based on the following formula:

[0080]

[0081] Where h represents the initial optical center position, hi represents the row number corresponding to the pixel, gi represents the gray value of the pixel, and k is the total number of pixels in the column direction.

[0082] S600: Extracts image data within a preset extraction range around the initial optical center position in the filtered laser line image.

[0083] After locating the initial optical center position, image data within a preset extraction range around the initial optical center position can be extracted. This extraction range can be set according to user needs, or it can be calculated based on the average width. Specifically, in some embodiments, the extraction range can be calculated based on the average width, extracting image data within the range above and below the initial optical center position from the filtered laser line image. The extraction range is defined as half the average width.

[0084] S700: Fit image data to obtain the laser line center.

[0085] After acquiring the image data, it can be fitted to a smooth curve to determine the laser line center. Various methods can be used to fit the image data, such as polynomial fitting and Gaussian curve fitting. Taking Gaussian curve fitting as an example, a Gaussian curve is obtained by fitting the image data. The laser line center is then extracted from the Gaussian curve, and the laser line center is the center of the Gaussian curve.

[0086] For example, such as Figure 4 As shown, after acquiring the filtered laser line image, and locating the initial optical center position within the image, image data within a range of k / 2 above and below the initial optical center position can be extracted, where k is the average width of the laser line calculated previously. Then, Gaussian curve fitting is performed on these image data to obtain the center of the Gaussian curve, which is the sub-pixel center value of the laser line (the optical center of the laser line).

[0087] Based on the above-described laser line optical center extraction method, some embodiments of this application also provide an apparatus for laser line optical center extraction, such as... Figure 5 As shown, Figure 5 This application provides a structural block diagram of a laser line optical center extraction method device. The edge-preserving filtering device includes an image acquisition unit, an image preprocessing unit, and a laser line optical center extraction unit, wherein:

[0088] The image acquisition unit is used to acquire images of the laser line to be extracted.

[0089] An image preprocessing unit is used to calculate the average width of the laser line in the laser line image to be extracted; calculate low-pass filtering parameters based on the average width; and perform low-pass filtering processing on the laser line image to be extracted according to the low-pass filtering parameters to obtain a filtered laser line image.

[0090] The laser line optical center extraction unit is used to locate the initial optical center position in the filtered laser line image, and to extract image data within a preset extraction range around the initial optical center position; and to fit the image data to obtain the laser line optical center.

[0091] In order to improve the speed of line optical center extraction, the image preprocessing unit and the laser line optical center extraction unit can be chips designed based on field-programmable gate array (FPGA).

[0092] In summary, this application provides a high-precision laser line optical center extraction method and system based on FPGA. The method, after acquiring an image of the laser line to be extracted, calculates the average width of the laser line in the image, calculates low-pass filtering parameters based on the average width, and performs low-pass filtering on the laser line image to obtain a filtered laser line image. It then locates the initial optical center position in the filtered laser line image and extracts image data within a preset extraction range around the initial optical center position. Finally, it fits the image data to obtain the laser line optical center. This method allows for preprocessing of the laser line image to be extracted, filtering out image noise, thereby ensuring the robustness and applicability of the subsequent optical center extraction algorithm, while improving the accuracy and speed of line optical center extraction.

[0093] Similar parts between the embodiments provided in this application can be referred to mutually. The specific implementation methods provided above are only a few examples under the overall concept of this application and do not constitute a limitation on the scope of protection of this application. For those skilled in the art, any other implementation methods extended from the solution of this application without creative effort shall fall within the scope of protection of this application.

Claims

1. A high-precision laser line center extraction method based on FPGA, characterized in that, include: Acquire the image of the laser line to be extracted; Calculate the average width of the laser lines in the laser line image to be extracted; Calculate the low-pass filter parameters based on the average width; The laser line image to be extracted is subjected to low-pass filtering processing according to the low-pass filtering parameters to obtain a filtered laser line image; Locate the initial optical center position in the filtered laser line image, and extract image data within a preset extraction range around the initial optical center position; The image data is fitted to obtain the laser line optical center; The method further includes: Calculate the cutoff frequency of the low-pass filter based on the average width; Image noise with a frequency greater than the cutoff frequency is filtered out from the laser line image to be extracted in order to obtain a filtered laser line image.

2. The laser line center extraction method according to claim 1, characterized in that, Calculating the average width of the laser lines in the laser line image to be extracted includes: The laser line image to be extracted is binarized to generate a binarized image; Calculate the average width of the laser lines in the binarized image.

3. The laser line center extraction method according to claim 2, characterized in that, The step of calculating the average width of the laser lines in the binarized image, the method includes: Traverse the pixels in the binarized image in the column direction; The number of consecutive pixels with a pixel value greater than zero in each column of the binarized image is calculated to obtain the laser line column width, which is the maximum number of consecutive pixels in the column. The laser line array widths are sorted in ascending or descending order according to their width values ​​to generate a width array. The average width is calculated based on the width sequence, where the average width is the median of the width sequence.

4. The laser line center extraction method according to claim 1, characterized in that, The method includes the step of performing low-pass filtering processing on the laser line image to be extracted according to the low-pass filtering parameters: Perform a Fast Fourier Transform on the laser line image to be extracted to generate a spectrum; Move the zero-frequency domain portion of the spectrum to the center of the spectrum; Set the Gaussian low-pass filter according to the low-pass filter parameters described above; The spectrum is subjected to low-pass filtering based on a Gaussian low-pass filter, and the zero-frequency domain portion is restored to its initial position after low-pass filtering. Perform an inverse fast Fourier transform on the spectrum to generate a filtered laser line image.

5. The laser line center extraction method according to claim 1, characterized in that, The method for locating the preliminary optical center position in the filtered laser line image includes: The preliminary optical center position is calculated based on the following formula: ； Where h represents the initial optical center position, h i g represents the row number corresponding to a pixel. i This represents the grayscale value of a pixel, where k is the total number of pixels in the column direction.

6. The laser line center extraction method according to claim 1, characterized in that, The method further includes the step of extracting image data within a preset range around the initial optical center position, and the method also includes: The extraction range is calculated based on the average width, and the extraction range is half of the average width. Image data within the extraction range above and below the initial optical center position is extracted from the filtered laser line image.

7. The laser line center extraction method according to claim 1, characterized in that, The step of fitting the image data, the method includes: Gaussian curve fitting is performed on the image data to obtain a Gaussian curve; The optical center of the laser line is extracted based on the Gaussian curve, where the optical center of the laser line is the center of the Gaussian curve.

8. A high-precision laser line center extraction system based on FPGA, characterized in that, include: The image acquisition unit is used to acquire images of the laser line to be extracted; An image preprocessing unit is used to calculate the average width of the laser line in the laser line image to be extracted; Calculate the low-pass filter parameters based on the average width; perform low-pass filtering on the laser line image to be extracted according to the low-pass filter parameters to obtain a filtered laser line image; A laser line optical center extraction unit is used to locate the initial optical center position in the filtered laser line image, and to extract image data within a preset extraction range around the initial optical center position; and to fit the image data to obtain the laser line optical center. The system is also used to calculate the cutoff frequency of the low-pass filter based on the average width; and to filter out image noise with a frequency greater than the cutoff frequency in the laser line image to be extracted, so as to obtain a filtered laser line image.

9. The laser line center extraction system according to claim 8, characterized in that, The image preprocessing unit is also used for: Obtain the pixel threshold; Traverse the pixels in the laser line image to be extracted in the column direction; The number of consecutive pixels in each column of the laser line image to be extracted that have a pixel value greater than the pixel threshold is calculated to obtain the laser line column width, which is the maximum number of consecutive pixels in the column. The laser line array widths are sorted in ascending or descending order according to their width values ​​to generate a width array. The average width is calculated based on the width sequence, where the average width is the median of the width sequence.

Images