Image processing method and device, electronic equipment and storage medium

Abstract

The present application provides an image processing method, device, equipment, storage medium and computer program product, the image processing method comprises: obtaining an image to be processed and N reference images, the image to be processed and the reference images have first noise respectively, N is a positive integer greater than or equal to 2;According to the N reference images, a correction data map is obtained, the correction data map includes the first noise after smoothing processing;The image to be processed and the correction data map are weighted and averaged to obtain a mean value map removing the first noise.The present application can carry out the noise reduction processing to the image, improves the accuracy of target information identification in the image.

Description

Technical Field

[0001] This invention relates to an image processing method, apparatus, electronic device, and storage medium, belonging to the field of image processing technology. Background Technology

[0002] Acquired images often contain noise, which affects the recognition of target information. Therefore, noise reduction processing of images to improve the accuracy of target information recognition is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0003] This invention provides an image processing method, apparatus, device, and storage medium that can perform noise reduction processing on images to improve the accuracy of target information recognition.

[0004] In one aspect, the present invention provides an image processing method, comprising:

[0005] Obtain the image to be processed and N reference images, wherein the image to be processed and the reference images each have a first noise, and N is a positive integer greater than or equal to 2;

[0006] A correction data map is obtained based on the N reference images, and the correction data map includes the first noise after smoothing.

[0007] The image to be processed is weighted and averaged with the correction data image to obtain the mean image after removing the first noise.

[0008] Optionally, the step of obtaining a correction data map based on the N reference images, wherein the correction image includes the first noise after smoothing, includes:

[0009] The average image of the N reference images is obtained to get a background image containing the first noise;

[0010] The background image is inverted to obtain the corrected data image.

[0011] Optionally, N > 10.

[0012] Optionally, the image to be processed is a fingerprint image; and / or, the first noise includes stripe noise, optionally, the stripe noise includes moiré patterns; and / or, the first noise includes periodic noise or non-periodic noise.

[0013] Optionally, it further includes: performing the smoothing process on the background image and then performing the color inversion process; and / or performing the smoothing process on the background image after the color inversion process to obtain the corrected data image.

[0014] Optionally, the first noise includes stripe noise, and the smoothing process is to smooth along the stripe noise lines.

[0015] Optionally, the step of obtaining the average image of the N reference images to obtain a background image containing the first noise includes: each of the N reference images has a one-to-one corresponding pixel with any other one; for each pixel, the average pixel value of that pixel in the N reference images is taken to obtain the average image; preferably, the pixel value is a grayscale value.

[0016] Optionally, when performing the weighted average, the weight ratio coefficient of the image to be processed is w, and the weight ratio coefficient of the correction image is 1-w, where 0.4≤w≤0.8.

[0017] Optionally, after obtaining the mean map with the first noise removed, the method further includes: performing contrast enhancement processing on the mean map; optionally, the contrast enhancement processing is performed using histogram equalization.

[0018] In another aspect, the present invention provides an image processing apparatus, comprising: an image acquisition module for acquiring an image to be processed and N reference images, wherein the image to be processed and the reference images each have a first noise, and N is a positive integer greater than or equal to 2; a correction data map acquisition module for obtaining a correction data map based on the N reference images, wherein the correction data map includes the first noise after smoothing; and a weighted averaging module for performing a weighted average of the image to be processed and the correction data map to obtain a mean map.

[0019] Optionally, the correction data image acquisition module includes: an averaging module, used to acquire the average image of the N reference images to obtain a background image containing the first noise; and a color inversion module, used to perform color inversion processing on the background image to obtain the correction data image.

[0020] Optionally, it also includes a smoothing module, used to smooth the background image before performing the color inversion processing, and / or to smooth the background image after the color inversion processing to obtain the corrected data image.

[0021] In another aspect, the present invention provides an electronic device including a processor, a memory, and a computer program, the computer program being stored in the memory and configured to be executed by the processor to implement the above-described image processing method.

[0022] In another aspect, the present invention provides a storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the image processing method described above.

[0023] The image processing method provided by this invention obtains a correction data map based on N reference images, including a first noise layer containing smoothing. The obtained correction data map is then weighted and averaged with the image to be processed to cancel out fixed noise in the image to be processed. This reduces or even eliminates fixed noise (such as fixed stripe noise, background noise, etc.) in the generated mean map, preventing these noises from interfering with target information in the image and improving the accuracy of target information recognition. Furthermore, the image processing method provided by this invention has advantages such as simple algorithm, reduced computational load, fast processing speed, and high efficiency, making it suitable for practical applications. Attached Figure Description

[0024] Figure 1 This is a flowchart of an image processing method according to an embodiment of the present invention;

[0025] Figure 2 This is a flowchart of an image processing method according to another embodiment of the present invention;

[0026] Figure 3 This is a flowchart of an image processing method according to another embodiment of the present invention;

[0027] Figure 4 This is an image to be processed in one embodiment of the present invention;

[0028] Figure 5 This is an average image in one embodiment of the present invention;

[0029] Figure 6 This is a background image that has been smoothed but not inverted in one embodiment of the present invention;

[0030] Figure 7 This is a correction data image generated by sequentially smoothing and inverting the background image in one embodiment of the present invention;

[0031] Figure 8 and Figure 9 This is a mean map generated by weighting the image to be processed and the correction data map according to different weights in one embodiment of the present invention.

[0032] Figure 10 This is a background image that has been inverted but not smoothed in one embodiment of the present invention;

[0033] Figure 11 This is a correction data image generated by sequentially inverting and smoothing a background image in one embodiment of the present invention. Detailed Implementation

[0034] Acquired images often contain noise that affects the recognition of target information (the foreground of the image). For example, during the acquisition of fingerprint images, due to inherent factors of the acquisition equipment (such as defects in the lighting system or circuitry), fixed noise with characteristics such as fixed position and amplification may be generated, such as regular periodic diagonal ridge interference noise (e.g., moiré patterns) or other non-periodic fixed noise. At the same time, in addition to the inherent factors of the acquisition equipment, random noise may also be generated due to external factors such as the acquisition environment. These noises, superimposed on the fingerprint ridges, affect the recognition of the fingerprint as the target information and easily lead to the appearance of false features.

[0035] Common methods for removing periodic noise include frequency domain filtering, which requires switching from the spatial domain to the frequency domain, removing noise in the frequency domain, and then switching back to the spatial domain. However, these existing noise reduction methods are not easy to locate noise, have limited noise reduction effects, and suffer from drawbacks such as high computational cost and long processing time.

[0036] Therefore, denoising images to improve the accuracy of target information recognition remains a pressing technical problem to be solved in this field.

[0037] Therefore, embodiments of the present invention provide an image processing method, with reference to... Figure 1 , Figure 2 and Figure 3 The image processing method includes:

[0038] S110. Obtain the image to be processed and N reference images. The image to be processed and the reference images each have a first noise, and N is a positive integer greater than or equal to 2.

[0039] It is understandable that N reference images are used to extract background images corresponding to the background data in the image to be processed. Therefore, the background data of these reference images corresponds to the background data of the image to be processed, that is, the background data of the two are basically the same. The first noise in the reference image is basically the same as the first noise in the image to be processed in terms of position, amplification and other features. Specifically, the pixels in the reference image and the image to be processed can correspond one-to-one, that is, any pixel in each reference image has a corresponding pixel at the same position in the image to be processed.

[0040] In practice, the image to be processed and N reference images can be acquired by the same device. The same device can be the same device, or it can be different devices, but the structure and components (such as light systems and circuits) of these devices are the same. As a result, the background data of the acquired images and the fixed noise generated by the device itself are basically the same.

[0041] The image to be processed has a foreground (i.e. target information), such as a fingerprint. The processing method of this embodiment removes noise from the image to facilitate the identification of the foreground.

[0042] The reference image may or may not have the foreground present in the image to be processed, or it may have the foreground present in the image to be processed (i.e., the reference image may be an image that is substantially identical to the image to be processed). For example, the foreground is a fingerprint, and the image to be processed and the reference image are fingerprint images, respectively. In practice, the same fingerprint acquisition device can be used to collect the fingerprints of a living organism to obtain the image to be processed and these reference images. Both the image to be processed and these reference images are original images that have been collected, and they all have the fingerprint as the foreground and substantially identical background data.

[0043] The aforementioned reference image and the image to be processed can specifically be 8-bit grayscale images, where the grayscale value of each pixel is 0-255. The grayscale value represents the brightness of the pixel, with 255 representing the brightest (white) and 0 representing the darkest (black).

[0044] Generally, the aforementioned image to be processed and reference image may each include fixed noise (i.e., noise with fixed characteristics such as position and amplification caused by factors of the acquisition device itself (such as defects in the lighting system or circuit structure), and random noise (i.e., random noise with non-fixed characteristics such as position caused by external factors such as the acquisition environment). The processing method of this embodiment can remove these fixed and random noises. Specifically, the first noise may include fixed noise.

[0045] In addition, the first noise can be periodic noise or irregular non-periodic noise. For example, during the acquisition of fingerprint images, regular periodic diagonal noise may be generated due to defects in the structure of the acquisition device's light system or circuit.

[0046] In addition, the first noise may include stripe noise, which may be fixed noise caused by factors of the acquisition device itself, such as diagonal stripe noise generated during the acquisition of fingerprint images due to defects in the structure of the acquisition device's light system or circuit.

[0047] In some specific embodiments, the above-mentioned stripe noise may include moiré patterns, which can be efficiently removed by the processing method of the present invention.

[0048] S120. Obtain a correction data map based on N reference images, the correction data map including the first noise after smoothing.

[0049] Specifically, the calibration data image contains a first noise, which has been smoothed to facilitate the subsequent weighted averaging of the image to be processed and the calibration data image, thereby canceling out the first noise in the image to be processed, making the background uniform, and improving the accuracy of target information recognition.

[0050] Specifically, a correction data map is obtained based on N reference images. The correction data map includes the first noise after smoothing, specifically including:

[0051] S1201. Obtain the average image of N reference images to get a background image containing the first noise.

[0052] Specifically, when acquiring images, information such as noise generated by the acquisition device itself and background data is fixed. However, information such as the foreground generated by the acquisition object (e.g., if the acquisition object is a living organism's fingerprint, the foreground is the fingerprint) and the location of random noise generated by external factors such as the environment is random and not fixed. By averaging N reference images, we can obtain an average image of these reference images. Random and uncertain information (such as foreground and random noise) can be removed, while the fixed information such as background data and fixed noise is retained. Thus, a background image containing background data and fixed noise, but basically without foreground and random noise, is extracted.

[0053] The aforementioned N reference images have the same background data and noise characteristics. Each of these reference images has a one-to-one correspondence with the pixels of any other reference image. That is, each pixel in each reference image has a corresponding pixel at the same position in any other reference image. For each pixel, the average pixel value of that pixel in the N reference images is taken to obtain the average image of these reference images.

[0054] For example, each of the N reference images has pixel 1, pixel 2, pixel 3, ..., pixel m (m equals the total number of pixels in each reference image). Pixel a in each reference image corresponds to pixel a in any other reference image (a = 1, 2, 3, ..., m). That is, pixel 1 in each reference image corresponds to pixel 1 in any other reference image, pixel 2 in each reference image corresponds to pixel 2 in any other reference image, ..., and pixel m in each reference image corresponds to pixel m in any other reference image. Calculate the average pixel value of pixel 1, pixel 2, pixel 3, ..., pixel m of N reference images respectively, to generate the average image of N reference images, which is the background image. In other words, the average image / background image has pixels that correspond one-to-one with the reference images. The pixel in the average image / background image that corresponds to pixel a in the reference image is pixel a', and the pixel value of pixel a' is the average pixel value of pixel a in the N reference images.

[0055] Specifically, the pixel value I of pixel a in N reference images can be calculated based on the following formula. a The average value I a' :

[0056] I a' =(I a1 +I a2 +I a3 +……+I aN ) / N

[0057] Among them, I a1 Let I be the pixel value of pixel a in the first reference image. a2 I is the pixel value of pixel a in the second reference image. a3 The pixel value of pixel a in the third reference image...I aN Let be the pixel value of pixel a in the Nth reference image, i.e. (I a1 +I a2 +I a3 +……+I aN ) represents the sum of pixel values ​​of pixel point a in N reference images; I a' It is the average pixel value of pixel a in N reference images (also the pixel value of pixel a' corresponding to pixel a in the average image / background image).

[0058] Specifically, the pixel value mentioned above can be a grayscale value, that is, the pixel value of pixel a is the grayscale value of pixel a.

[0059] Relatively speaking, the larger N is, the more reference images there are, which is more conducive to the accuracy of the extracted background image data. That is, it improves the matching degree between its background data / fixed noise and other information and the image to be processed, and can further remove random noise and improve the smoothness of the background image. In some specific embodiments, N > 10.

[0060] S1202. Invert the background image to obtain the correction data image.

[0061] Specifically, after the background image undergoes color inversion processing, its bright and dark areas are reversed (i.e., bright areas become dark areas, and dark areas become bright areas). For example, before color inversion processing, the grayscale value of a certain pixel 'a' in the background image is 255. After color inversion processing, the grayscale value of the corresponding pixel 'a' in the generated correction data image becomes 0. Before color inversion processing, the grayscale value of a certain pixel 'b' in the background image is 0. After color inversion processing, the grayscale value of the same pixel 'b' in the generated correction data image is 255.

[0062] Therefore, by inverting colors, the bright and dark areas in the calibration data image are reversed compared to those in the image to be processed. For example, if a pixel 'a' in the image to be processed is a bright area, the corresponding pixel in the calibration data image is a dark area; similarly, if a pixel 'b' in the image to be processed is a dark area, the corresponding pixel in the calibration data image is a bright area. In subsequent step S1203, a weighted average is applied between the image to be processed and the calibration data image, canceling out the brightness and darkness of each pixel / position, thereby removing noise (especially fixed noise), making the background uniform, obtaining a denoised mean image, and improving the accuracy of target information recognition.

[0063] Optionally, such as Figure 2 and Figure 3 As shown, the above image processing method may further include S1203, smoothing the background image.

[0064] Generally, when N is large, the smoothness of the obtained background image is high, and step S1203 can be omitted. When N is small, the smoothness of the obtained background image is poor, and step S1203 can be performed to further reduce noise in the background image.

[0065] In addition, such as Figure 2 As shown, the background image can be smoothed before color inversion (i.e., step S1203 is executed first, followed by step S1202).

[0066] Or, such as Figure 3As shown, after the background image is inverted, it is then smoothed. That is, the background image after inversion is smoothed (step S1202 is executed first, then step S1203 is executed) to generate a correction data image.

[0067] Alternatively, as needed, the background image obtained in step S1201 can be processed by smoothing, inverting colors, and smoothing again in sequence (i.e., by performing steps S1203, S1202, and S1203 in sequence) to generate a correction data image.

[0068] In some embodiments, the first noise includes stripe noise (i.e., the background image contains stripe noise), and the smoothing process can specifically be smoothing along the lines of the stripe noise. After the smoothing process, the direction of the lines of these stripe noises remains basically unchanged.

[0069] In practice, at least one of the following smoothing methods can be used for the above smoothing process: mean smoothing, Gaussian smoothing, median smoothing, and Laplace smoothing.

[0070] S130. Perform a weighted average of the image to be processed and the correction data image to obtain the mean image after removing the first noise.

[0071] As described above, the color inversion process in step S1202 reverses the bright and dark areas in the correction data map and the image to be processed. By weighting the image to be processed and the correction data map, the brightness and darkness of each pixel / position are canceled out, thereby removing interference noise such as the first noise, making the background uniform, obtaining the denoised mean map, and improving the recognition accuracy of target information.

[0072] Specifically, the calibrated data image and the image to be processed have a one-to-one correspondence of pixels. That is, each pixel in the calibrated data image has a corresponding pixel at the same position in the image to be processed. For each pixel, the pixel value of the pixel in the two images (i.e., the image to be processed and the data calibrated image) is weighted and averaged (i.e., the weighted average value of the pixel in the two images is calculated according to the preset weight ratio coefficient of the two images) to obtain a weighted average image (i.e., the mean image) of the image to be processed and the calibrated data image.

[0073] For example, both the calibration data image and the image to be processed have pixels 1', 2', 3', ..., m' (m' equals the total number of pixels in either the calibration image or the image to be processed). Pixel c in the calibration data image corresponds to pixel c in the image to be processed (c = 1', 2', 3', ..., m'). That is, pixel 1' in the calibration data image corresponds to pixel 1' in the image to be processed, pixel 2' in the calibration data image corresponds to pixel 2' in the image to be processed, ..., and pixel m' in the calibration data image corresponds to pixel m' in the image to be processed. The pixel value of pixel 1' in the calibration data image and the pixel value of the image to be processed are calculated respectively. The weighted average of pixel values ​​of pixel 1' in the processed image, the weighted average of pixel values ​​of pixel 2' in the calibration data image and pixel values ​​of pixel 2' in the image to be processed, ..., the weighted average of pixel values ​​of pixel m' in the calibration data image and pixel values ​​of pixel m' in the image to be processed, are used to generate a weighted average image (i.e., a mean image) of the image to be processed and the calibration data image. That is, the mean image has pixels that correspond one-to-one with the calibration data image / image to be processed. The pixel in the mean image corresponding to pixel c in the calibration data image / image to be processed is pixel c', and the pixel value of pixel c' is the pixel value V of pixel c in the calibration data image. c2 The pixel value V of pixel c in the image to be processed c1 The weighted average V c' .

[0074] Specifically, the pixel value V of pixel c in the correction data image can be calculated based on the following formula. c2 The pixel value V of pixel c in the image to be processed c1 The weighted average V c' :

[0075] V c' =w×V c1 +(1-w)×V c2

[0076] Where 0 < w < 1, w is the weight ratio coefficient of the image to be processed, and 1-w is the weight ratio coefficient of the correction data map.

[0077] For example, w can be 0.4, 0.5, 0.6, 0.7, 0.8, 0.9 or a range consisting of any two of them.

[0078] In some preferred embodiments, 0.4≤w≤0.8, and by controlling the weights of the image to be processed and the image to be corrected within such a range, the noise reduction effect can be further improved.

[0079] The pixel values ​​mentioned above can specifically be grayscale values. For example, the pixel value of pixel a is the grayscale value of pixel a, and the pixel value of pixel c is the grayscale value of pixel c.

[0080] Furthermore, the image processing method of this embodiment may also include S140, performing contrast enhancement processing on the mean image to obtain the final denoised image.

[0081] Specifically, by performing contrast enhancement processing on the mean image, the contrast of the mean image can be enhanced, further improving the accuracy of target information (such as fingerprints) recognition in the obtained denoised image. Subsequently, the denoised image can be output, completing the image processing operation.

[0082] For example, the mean plot can be enhanced for contrast using methods such as histogram equalization.

[0083] It should be noted that in the embodiments of the present invention, the image to be processed and the N reference images can be acquired simultaneously or not simultaneously. For example, the N reference images can be acquired first through step S110, and after executing steps S1201, S1202 and S1203, the image to be processed can be acquired in the manner of step S110, and then subsequent steps such as S130 can be executed.

[0084] The following section uses fingerprint image processing as an example to illustrate... Figure 2 The process shown and Figure 3 The specific embodiments of the process shown further illustrate the image processing method of this application.

[0085] Figure 2 This is a flowchart of an image processing method according to an embodiment of the present invention, with reference to... Figure 2 The steps S110, S1201, S1203, S1202, S130, and S140 are executed sequentially. The specific process of each step is as described above and will not be repeated here. The image processing method of this embodiment is used to denoise the fingerprint image. The image to be processed and N (N>10) reference images are fingerprint images, specifically 8-bit grayscale images, with the fingerprint as the foreground and the first noise being diagonal / moiré patterns (e.g.,...). Figures 4 to 7 (As shown).

[0086] in, Figure 4 To perform the process of step S110, the image to be processed (similar to the reference image) is obtained. Figure 5 The average image obtained in step S1201 (i.e., the background image without smoothing and color inversion) Figure 6 This is the background image after smoothing (step S1203). Figure 7 This is the correction data image generated after inverting the background image (step S1202). Figure 8 and Figure 9 These are the mean images obtained by weighting the image to be processed and the corrected data image according to different weights (step S130). Figure 8 It is the mean image after weighting the images to be processed according to the weight ratio coefficient w = 0.7. Figure 9 It is the mean image after weighting the images to be processed according to the weight ratio coefficient w = 0.9.

[0087] from Figure 4 As can be seen, fingerprint images contain fingerprints and noise, including diagonal ridge noise and random noise.

[0088] from Figure 5 It can be seen that after taking the average image of N reference images in step S1201, fingerprints and random noise are removed, while background and diagonal noise are retained.

[0089] from Figure 6 As can be seen, after the smoothing process in step S1203, random noise is further removed, and the smoothness of the background image is improved.

[0090] from Figure 7 As can be seen, after color inversion, the bright and dark areas of the image are reversed. That is, the dark areas in the uninverted background image become bright areas, and the bright areas become dark areas after color inversion. For example, combined with... Figures 4 to 6 The central portion of the image to be processed (original image) and the uninverted background image is darker (lower grayscale), while the central portion of the generated correction data image is brighter (higher grayscale) after the background image is inverted. Therefore, by weighted averaging the image to be processed and the correction data image, the brightness and darkness are canceled out, resulting in a mean image with a uniform background and removed diagonal noise. Figure 8 or Figure 9 ).

[0091] Combination Figure 4 , Figure 8 and Figure 9 As can be seen, the image processing method in this embodiment effectively removes random noise and fixed noise (diagonal stripes) from the image to be processed.

[0092] Combination Figure 8 (w=0.7) and Figure 9 (w=0.9) It can be seen that, Figure 8 The background is more uniform, and the fingerprints are clearer, which can further improve the accuracy of fingerprint recognition.

[0093] Figure 3 This is a flowchart of an image processing method according to another embodiment of the present invention, and... Figure 2The only difference in the embodiment shown is that step S1202 is executed first, followed by step S1203 (the background image obtained in step S1201 is first inverted and then smoothed to generate a correction data image), that is, steps S110, S1201, S1202, S1203, S130, and S140 are executed in sequence. Figure 10 This refers to the background image obtained in step S1201 that has undergone color inversion processing in step S1202, but has not undergone smoothing processing in step S1203. Figure 11 This is the background image after the background image has undergone color inversion processing in step S1202 and smoothing processing in step S1203.

[0094] Combination Figure 2 In the illustrated embodiment Figures 5 to 7 ,as well as Figure 3 In the illustrated embodiment Figure 10 and Figure 11 As can be seen, smoothing the background image before inverting the colors, or inverting the background image before smoothing, can effectively remove random noise from the background image and obtain a corrected data image with uniform background noise containing diagonal stripes.

[0095] In addition, combined Figure 5 and Figure 6 , Figure 10 and Figure 11 It can be seen that when step S1203 is executed, the lines of the diagonal lines are smoothed. After the smoothing process, the direction of these diagonal lines remains basically unchanged.

[0096] This invention also provides an image processing apparatus, comprising: an image acquisition module for acquiring the image to be processed and N reference images; a correction data map acquisition module for obtaining the correction data map based on the N reference images; and a weighted average module for performing a weighted average of the image to be processed and the correction data map to obtain a mean map.

[0097] Specifically, the calibration data image acquisition module may include: an averaging module for acquiring the average image of N reference images to obtain a background image; and a color inversion module for inverting the background image to obtain the calibration data image.

[0098] In addition, the above-mentioned correction data image acquisition module may also include a smoothing module for smoothing the average image to obtain a background image, and / or for smoothing the background image after color inversion to obtain a correction data image.

[0099] The image processing apparatus provided in this embodiment of the invention can be used to implement the above-described image processing method. Its implementation principle is similar to that of the above-described image processing method, and it has the same excellent effect as the above-described image processing method, which will not be described in detail here.

[0100] This invention also provides an electronic device, including a processor, a memory, and a computer program, wherein the computer program is stored in the memory and configured to be executed by the processor to implement the above-described image processing method.

[0101] This invention also provides a storage medium storing a computer program, which is executed by a processor to implement the image processing method described above. The storage medium is a computer-readable storage medium, such as a memory containing instructions (computer program) that can be executed by the processor of the electronic device to complete the image processing method. For example, the storage medium is a non-transitory computer-readable storage medium, such as a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, or optical data storage device.

[0102] This invention also provides a computer program product, including a computer program, which is executed by a processor to implement the above-described image processing method.

[0103] According to some embodiments of the present invention, the above-described electronic device includes at least one processor, which can read a computer program from a readable storage medium, and the at least one processor executes the computer program to cause the electronic device to perform the above-described image processing method.

[0104] 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 scope of the technical solutions of the embodiments of the present invention.

Claims

1. An image processing method, characterized in that, include: The process involves acquiring a target image and N reference images. The target image and the reference images each possess a first noise level. N is a positive integer greater than 10. The background data of the N reference images corresponds to the background data of the target image, and each pixel in the N reference images corresponds one-to-one with the target image. The first noise is a fixed noise with fixed characteristics generated by the acquisition device itself, and this first noise is stripe noise. The target image and the N reference images are acquired using the same device, and the structure and components of this same device are identical. Each of the N reference images has a one-to-one corresponding pixel with any other one. For each pixel, the average pixel value of that pixel in the N reference images is taken to obtain the average image of the N reference images, so as to obtain a background image containing only the first noise and excluding foreground and random noise. The background image is inverted, and the background image after inversion is smoothed along the ridge direction of the stripe noise to obtain a correction data image. The correction data image includes the first noise that is opposite in brightness to the image to be processed and has been smoothed. The image to be processed is weighted and averaged with the correction data image. The first noise in the image to be processed is eliminated by canceling out the brightness of the images, resulting in a mean image with the first noise removed.

2. The image processing method according to claim 1, characterized in that, The image to be processed is a fingerprint image; the stripe noise includes moiré patterns.

3. The image processing method according to claim 1, characterized in that, The pixel value is a grayscale value.

4. The image processing method according to claim 1, characterized in that, When performing the weighted average, the weight ratio coefficient of the image to be processed is w, and the weight ratio coefficient of the corrected data image is 1-w, where 0.4≤w≤0.

8.

5. The image processing method according to claim 1, characterized in that, After obtaining the mean map with the first noise removed, the method further includes: performing contrast enhancement processing on the mean map; optionally, histogram equalization is used to perform the contrast enhancement processing.

6. An image processing apparatus, characterized in that, include: An image acquisition module is used to acquire an image to be processed and N reference images. The image to be processed and the reference images each have a first noise. N is a positive integer greater than 10. The background data of the N reference images corresponds to the background data of the image to be processed. Each pixel in the N reference images corresponds one-to-one with the pixel in the image to be processed. The first noise is a fixed noise with fixed characteristics generated by the acquisition device itself, and the first noise is stripe noise. The image to be processed and the N reference images are acquired using the same device, and the structure and components of the same device are identical. A calibration data map acquisition module is used to obtain a calibration data map based on the N reference images, wherein the calibration data map includes the first noise after smoothing. The weighted average module is used to perform a weighted average of the image to be processed and the correction data map, and eliminate the first noise in the image to be processed by canceling out the brightness and darkness to obtain the mean map. The calibration data map acquisition module includes: An averaging module is used to ensure that each of the N reference images has a one-to-one correspondence with any other pixel. For each pixel, the average pixel value of that pixel in the N reference images is taken to obtain an average image of the N reference images, so as to obtain a background image containing only the first noise and excluding foreground and random noise; the first noise includes stripe noise, and the smoothing process is to smooth along the stripe noise lines. The color inversion processing module is used to invert the background image and smooth the background image after color inversion along the ridge direction of the stripe noise to obtain the correction data image. The correction data image includes the first noise that is opposite in brightness to the image to be processed and has been smoothed.

7. An electronic device, characterized in that, It includes a processor, a memory, and a computer program, the computer program being stored in the memory and configured to be executed by the processor to implement the image processing method as described in any one of claims 1-5.

8. A storage medium, characterized in that, It stores a computer program, which is executed by a processor to implement the image processing method as described in any one of claims 1-5.

Images