Wavelet transform-based line array camera stripe noise removal method and device

Abstract

The application relates to the technical field of digital image processing, and discloses a line array camera stripe noise removing method and device based on wavelet transform, which comprises the following steps: acquiring image data of each row of a line array camera; performing wavelet transform on the image data of each row to eliminate response differences between pixels, so as to obtain calibrated image data of each row; splicing the calibrated image data of each row to obtain a complete image; and performing smoothing processing on the complete image to obtain a complete image without stripe noise, so that stripe noise caused by inconsistent responses of the line array camera can be effectively removed, and a better denoising effect can be achieved.

Description

Technical Field

[0001] This invention relates to the field of digital image processing technology, and in particular to a method and apparatus for removing stripe noise from a linear array camera based on wavelet transform. Background Technology

[0002] The working principle of a line scan camera is to acquire only one line of images with each exposure. After each line is acquired, the object being detected moves to the next line, at which point the line scan camera exposes again to acquire the second line of images, and so on, completing the scanning of the object being detected line by line. Then, the images of each line are stitched together to obtain a complete image of the object being detected.

[0003] A linear array camera's detector consists of photosensitive pixels arranged in a uniform "I" shape. Each pixel is responsible for receiving light signals from the image and converting them into electrical signals. However, due to minute differences in the size and material doping of each pixel during manufacturing, their photosensitive and response characteristics are not entirely consistent. This inconsistency manifests as stripe noise throughout the image. These stripes typically appear along the vertical or horizontal direction of the image, affecting image quality and accuracy. To eliminate or reduce this stripe noise, appropriate signal processing and image processing techniques are needed to ensure that the image data acquired from the linear array camera accurately reflects the true characteristics and state of the detected object.

[0004] Currently, direct hardware-based removal of stripe noise in images captured by line scan cameras is quite difficult because stripe noise is usually caused by minute differences in each row of pixels, which cannot be completely eliminated by simple hardware adjustments or corrections. Therefore, the common solution is to perform denoising processing on the 2D image at the software level. While this method can effectively reduce stripe noise, it is highly complex, requiring sophisticated algorithms and processing steps, and the processing time is relatively long.

[0005] Therefore, there is an urgent need for a method for removing stripe noise from line scan cameras based on wavelet transform, which can improve existing methods for denoising images captured by line scan cameras, adapt to the working characteristics and application scenarios of line scan cameras, and effectively remove stripe noise caused by inconsistencies in the response of line array detectors. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method and apparatus for removing stripe noise from a linear array camera based on wavelet transform, which can effectively remove stripe noise caused by inconsistencies in the response of the linear array detector, achieving better noise reduction results.

[0007] This invention provides a method and apparatus for removing stripe noise from a linear array camera based on wavelet transform, comprising the following steps: S1. Acquire image data for each row of the line scan camera.

[0008] Furthermore, each row of image data is acquired by a linear array detector in the linear array camera, and the linear array detector outputs one row of image data in each working cycle.

[0009] S2. By performing wavelet transform on each row of image data, the response differences between pixels are eliminated, and the calibrated image data for each row is obtained.

[0010] Furthermore, S21, the thresholds for preset wavelet basis functions, wavelet decomposition levels, and refinement coefficients.

[0011] Furthermore, the wavelet basis function is the Haar wavelet.

[0012] S22. Based on the wavelet basis function, perform wavelet decomposition on each row of image data to obtain approximation coefficients and refinement coefficients.

[0013] S23. Perform threshold processing on the fine coefficients; If the absolute value of the refinement coefficient is less than the threshold, then the refinement coefficient is set to zero; If the absolute value of the refinement coefficient is greater than or equal to the threshold, then the refinement coefficient is determined to be negative or non-negative, and the refinement coefficient is adjusted accordingly. When the refinement coefficient is non-negative, subtract the threshold value of the refinement coefficient from the refinement coefficient; When the refinement coefficient is negative, add the threshold value of the refinement coefficient to the refinement coefficient; S24. Continue to perform wavelet decomposition on the approximation coefficients; S25. Repeat steps S23 to S24 until the wavelet decomposition level is reached; S26. Perform wavelet inverse transform on all the approximation coefficients of all levels and the fine coefficients after thresholding to obtain the calibrated image data for each row.

[0014] S3. The calibrated image data of each row is stitched together to obtain a complete image.

[0015] Furthermore, based on the acquisition order of each row of data from the line scan camera, the calibrated image data of each row is stitched together sequentially to obtain a complete image.

[0016] S4. Smooth the complete image to obtain a complete image without stripe noise.

[0017] Furthermore, by applying a mean filtering algorithm to smooth the complete image, a complete image without stripe noise is obtained.

[0018] The present invention also provides a wavelet transform-based stripe noise removal device for linear array cameras, characterized in that it comprises: Image data acquisition module: used to acquire image data for each row of the line scan camera; Calibration module: By performing wavelet transform on each row of image data, the response differences between pixels are eliminated, and calibrated image data for each row is obtained; Image stitching module: connected to the calibration module, used to stitch together each row of calibrated image data to obtain a complete image; Smoothing module: Connected to the image stitching module, it is used to smooth the complete image to obtain a complete image without stripe noise.

[0019] The embodiments of the present invention have the following technical effects: This invention combines wavelet transform of one-dimensional signals with smoothing processing of two-dimensional images, aiming to effectively remove stripe noise caused by inconsistencies in the response of linear detectors.

[0020] By using wavelet transform of a one-dimensional signal, frequency domain analysis and processing can be performed on each line of image data acquired by a line scan camera in order to capture and reduce high-frequency components in stripe noise. This step can effectively eliminate the stripe effect caused by inconsistencies in the response of photosensitive pixels during the acquisition process of the line scan camera.

[0021] By smoothing the two-dimensional image and performing auxiliary homogenization, this processing method helps to further reduce any residual stripe noise in the image and improve the overall image quality and clarity.

[0022] Compared to traditional methods, this invention combines one-dimensional correction and two-dimensional smoothing when dealing with stripe noise in line scan cameras. This effectively addresses the stripe noise problem commonly encountered by line scan cameras during image acquisition, significantly shortening the denoising processing time, reducing algorithm complexity, improving image processing efficiency and accuracy, and achieving better denoising results. It is suitable for various application scenarios requiring high-quality image data. Attached Figure Description

[0023] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a flowchart of a method for removing stripe noise from a linear array camera based on wavelet transform, provided by an embodiment of the present invention. Figure 2 This is a schematic diagram of calibrated image data stitching provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the experimental results of a wavelet transform-based method for removing stripe noise from a linear array camera, provided in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0026] Figure 1 This is a flowchart illustrating a wavelet transform-based method for removing stripe noise from a linear array camera, as provided in an embodiment of the present invention. See also... Figure 1 Specifically, it includes: S1. Acquire image data for each row of the line scan camera; Image data for each row is acquired by a linear array detector in a linear array camera, and the linear array detector outputs one row of image data in each working cycle.

[0027] Specifically, the linear detector in the linear scan camera outputs one line of image data per working cycle. ,in This represents the image data in row m, where M is the total number of rows captured by the linear scan camera. This represents the number of pixels in each row of image data.

[0028] For example, the line scan camera outputs a total of 512 lines of image data.

[0029] S2. By performing wavelet transform on each row of image data, the response differences between pixels are eliminated, and the calibrated image data for each row is obtained. S21. Preset thresholds for wavelet basis functions, wavelet decomposition levels, and refinement coefficients; The wavelet basis function is the Haar wavelet; The formula for the wavelet transform is as follows: ; ; in, The scaling function for the Haar wavelet; The wavelet function of the Haar wavelet; , j is the scaling parameter, representing the scaling function and wavelet function in the frequency domain, and J-1 is the maximum level of wavelet decomposition; k is the translation parameter, representing the translation of the scaling function and the wavelet function in the time domain; These are approximate coefficients. For refinement coefficients; This is the starting level for wavelet decomposition.

[0030] For example, for ease of description, the approximation coefficient and the refinement coefficient are denoted as C. i and D i ,i represents the level of wavelet decomposition; The threshold T is determined using the sqtwolog principle, as shown in the following formula:

[0031] in, l The refinement factor D for the current level i The length.

[0032] For example, for the first-level wavelet decomposition, the length of D1 is 256, and the threshold T in the first-level wavelet decomposition is 3.3302. Similarly, the thresholds corresponding to other levels of decomposition can be obtained.

[0033] When performing wavelet decomposition on 512 rows of image data using Haar wavelets, a maximum of 8 decompositions can be performed; therefore, the wavelet decomposition level L is 8. Each decomposition generates a specific number of low-frequency and high-frequency coefficients to represent different frequency components of the original image. This decomposition process helps in analyzing and compressing image data while preserving important visual information.

[0034] S22. Based on the wavelet basis function, perform wavelet decomposition on each row of image data to obtain approximation coefficients and refinement coefficients; S23. Perform threshold processing on the fine coefficients; If the absolute value of the refinement coefficient is less than the threshold, then the refinement coefficient is set to zero; If the absolute value of the refinement coefficient is greater than or equal to the threshold, then the refinement coefficient is determined to be negative or non-negative, and the refinement coefficient is adjusted accordingly. When the refinement coefficient is non-negative, subtract the threshold value of the refinement coefficient from the refinement coefficient; When the refinement coefficient is negative, add a threshold value to the refinement coefficient; specifically, use a soft thresholding principle to refine the refinement coefficient D. i The processing is performed using the following formula: ; in, These are the refined coefficients after thresholding. When D i When the absolute value is not less than the threshold T, then according to D i The symbol for D i Make adjustments; when D i When the value is non-negative, the refinement coefficient is subtracted from the threshold, and the result is: =D i -T; when D i When the value is negative, the refinement coefficient is added to the threshold, and the result is: =D i +T; when D i When the absolute value of D is less than the threshold T, D will be... i Set to zero, that is =0.

[0035] S24. Continue to perform wavelet decomposition on the approximation coefficients; S25. Repeat steps S23 to S24 until the wavelet decomposition level is reached; For example, the refinement coefficients decomposed at each level and approximation coefficients Perform steps S23-S24 on all rows, where i=1,2,...,8, to eliminate each row of image data. Inter-pixel response differences.

[0036] S25. Perform wavelet inverse transform on all the approximation coefficients of all levels and the refined coefficients after thresholding to obtain the calibrated image data for each row.

[0037] Specifically, after completing the L-level wavelet decomposition, the approximation coefficients of all levels and the refined coefficients after thresholding are subjected to inverse wavelet transform to obtain the calibrated image data for each row. The formula for inverse wavelet transform is as follows:

[0038] in, For each row of image data after calibration; The scaling function for the Haar wavelet; The wavelet function of the Haar wavelet; , j is the scaling parameter, representing the scaling function and wavelet function in the frequency domain, and J-1 is the maximum level of wavelet decomposition; k is the translation parameter, representing the translation of the scaling function and the wavelet function in the time domain; These are approximate coefficients. For refinement coefficients; This is the starting level for wavelet decomposition.

[0039] S3. The calibrated image data of each row is stitched together to obtain a complete image; Based on the acquisition order of each row of data from the line scan camera, the calibrated image data of each row is stitched together to obtain a complete image.

[0040] Specifically, the linear detector in a linear camera outputs only one line of image per working cycle, but in actual use, the images are stitched together in sequence to obtain a complete image.

[0041] For example, such as Figure 2 As shown, the calibrated image data for each row... By stitching the images together in the order they were acquired, a complete image can be obtained. .

[0042] S4. Smooth the complete image to obtain a complete image without stripe noise; By applying the mean filtering algorithm to smooth the complete image, a complete image without stripe noise is obtained.

[0043] Since the aforementioned wavelet transform process is performed row by row, the complete image... Local non-uniformity may occur, thus requiring the use of smoothing algorithms to smooth the entire image. Perform smoothing processing.

[0044] In this step, mean filtering is used to smooth the stitched image, using the following convolution kernel:

[0045] Full image A certain pixel Its 3×3 neighborhood is denoted as ,but The filtered value is: ; For the complete image Each pixel in the image is sequentially subjected to the above smoothing filtering process to obtain a complete image free of stripe noise. .

[0046] This completes the removal of stripe noise from the linear scan camera. This refers to the image after stripe noise has been removed.

[0047] Figure 3 This is a schematic diagram illustrating the experimental results of a wavelet transform-based method for removing stripe noise from a linear array camera, as provided in an embodiment of the present invention. See also... Figure 3 As shown, it can be observed that the original image 3(a) contains a large amount of stripe noise. After wavelet transform, the noise in each row of image data is eliminated. The response differences between pixels are used to obtain the calibrated image data for each row. The calibrated image data of each row are then stitched together in sequence to obtain a complete image 3(b) that no longer contains stripe noise. The complete image 3(b) is then smoothed to obtain a complete image 3(c) that does not contain stripe noise and does not contain non-uniformity information.

[0048] In summary, based on the working characteristics of line scan cameras, this invention utilizes one-dimensional wavelet transform to remove stripe noise from each row of image data, and then combines it with two-dimensional image smoothing processing to complete the removal of stripe noise from line scan cameras. This reduces the complexity of the stripe noise removal algorithm, shortens the processing time, and achieves better stripe noise removal results compared to existing technologies that directly denoise two-dimensional images.

[0049] The present invention also provides a wavelet transform-based stripe noise removal device for linear array cameras, characterized in that it comprises: Image data acquisition module: used to acquire image data for each row of the line scan camera; Calibration module: By performing wavelet transform on each row of image data, the response differences between pixels are eliminated, and calibrated image data for each row is obtained; Image stitching module: connected to the calibration module, used to stitch together each row of calibrated image data to obtain a complete image; Smoothing module: Connected to the image stitching module, it is used to smooth the complete image to obtain a complete image without stripe noise.

[0050] It should be noted that the terminology used in this invention is for describing specific embodiments only and is not intended to limit the scope of this application. As shown in this specification, unless the context clearly indicates otherwise, words such as "a," "an," "an," and / or "the" do not specifically refer to the singular and may include the plural. The terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, or apparatus that includes said element.

[0051] It should also be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Unless otherwise expressly specified and limited, the terms "installed," "connected," "linked," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components. For those skilled in the art, the specific meaning of the above terms in the present invention can be understood according to the specific circumstances.

[0052] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A method for removing stripe noise from a linear array camera based on wavelet transform, characterized in that, Includes the following steps: S1. Acquire image data for each row of the line scan camera; S2. By performing wavelet transform on each row of image data, the response differences between pixels are eliminated, and the calibrated image data for each row is obtained. S3. The calibrated image data of each row is stitched together to obtain a complete image; S4. Smooth the complete image to obtain a complete image without stripe noise; By performing wavelet transform on each row of image data to eliminate inter-pixel response differences, calibrated image data for each row is obtained, specifically including: S21. Preset thresholds for wavelet basis functions, wavelet decomposition levels, and refinement coefficients; S22. Based on the wavelet basis function, perform wavelet decomposition on each row of image data to obtain approximation coefficients and refinement coefficients; S23. Perform threshold processing on the fine coefficients; If the absolute value of the refinement coefficient is less than the threshold value of the refinement coefficient, then the refinement coefficient is set to zero; If the absolute value of the fineness coefficient is not less than the threshold of the fineness coefficient, then it is determined whether the fineness coefficient is negative or non-negative, and the fineness coefficient is adjusted accordingly. When the refinement coefficient is non-negative, subtract the threshold value of the refinement coefficient from the refinement coefficient; When the refinement coefficient is negative, add the threshold value of the refinement coefficient to the refinement coefficient; S24. Continue to perform wavelet decomposition on the approximation coefficients; S25. Repeat steps S23 to S24 until the wavelet decomposition level is reached; S26. Perform wavelet inverse transform on all the approximation coefficients of all levels and the fine coefficients after thresholding to obtain the calibrated image data for each row.

2. The method for removing fringe noise from a linear array camera based on wavelet transform according to claim 1, characterized in that, The acquisition of each row of image data from the line scan camera specifically includes: Image data for each row is acquired by a linear array detector in a linear array camera, and the linear array detector outputs one row of image data in each working cycle.

3. The method for removing stripe noise from a linear array camera based on wavelet transform according to claim 1, characterized in that, The step of stitching together the calibrated image data from each row to obtain a complete image specifically includes: Based on the acquisition order of each row of data from the line scan camera, the calibrated image data of each row is stitched together to obtain a complete image.

4. The method for removing stripe noise from a linear array camera based on wavelet transform according to claim 1, characterized in that, The wavelet basis function is the Haar wavelet.

5. The method for removing fringe noise from a linear array camera based on wavelet transform according to claim 1, characterized in that, In step S4, smoothing the complete image to obtain a complete image without stripe noise specifically includes: By applying the mean filtering algorithm to smooth the complete image, a complete image without stripe noise is obtained.

6. A device for removing stripe noise from a linear array camera based on wavelet transform, characterized in that, The method for removing stripe noise from a linear array camera based on wavelet transform, as described in any one of claims 1-5, is characterized by comprising: Image data acquisition module: used to acquire image data for each row of the line scan camera; Calibration module: By performing wavelet transform on each row of image data, the response differences between pixels are eliminated, and calibrated image data for each row is obtained; Image stitching module: connected to the calibration module, used to stitch together each row of calibrated image data to obtain a complete image; Smoothing module: Connected to the image stitching module, it is used to smooth the complete image to obtain a complete image without stripe noise.

Images