Image feature point screening method and device and electronic equipment

Abstract

The embodiment of the invention discloses an image feature point screening method and device and electronic equipment. A specific embodiment of the method comprises the steps of obtaining a to-be-processed image and a noise parameter of the to-be-processed image; according to the noise parameters, variance stabilization transformation is carried out on the to-be-processed image, a transformed image is obtained, and the gray value range of the transformed image is different from that of the to-be-processed image; normalizing the gray value range of the transformed image to the gray value range of the to-be-processed image, and determining a normalization multiplying power; taking the normalized multiplying power as a target noise standard deviation, and determining a feature point response value corresponding to the target noise standard deviation as a feature point screening threshold value of the to-be-processed image; and screening the feature points of the to-be-processed image by using the feature point screening threshold. According to the embodiment, the extracted feature points can be reasonably screened, so that the high-quality feature points are reserved.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of computer technology, and more particularly to a method, device, and electronic device for screening image feature points. Background Art

[0002] Existing feature point extraction algorithms, such as Harris and Fast, typically use a globally fixed threshold across the entire image. Depending on the image content and the noise level of the current image detector, false feature points may be detected in areas that are typically considered flat.

[0003] Incorrect feature detection results will lead to errors or failures in subsequent feature matching and depth filtering modules, which in turn leads to degradation of positioning accuracy. Therefore, how to extract high-quality feature points is a technical problem that needs to be solved urgently. Summary of the Invention

[0004] This disclosure section is provided to briefly introduce concepts that will be described in detail in the detailed description section below. This disclosure section is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0005] In a first aspect, an embodiment of the present disclosure provides an image feature point screening method, comprising: obtaining an image to be processed and noise parameters of the image to be processed; performing a variance stabilization transformation on the image to be processed according to the noise parameters to obtain a transformed image, wherein the grayscale value range of the transformed image is different from the grayscale value range of the image to be processed; normalizing the grayscale value range of the transformed image to the grayscale value range of the image to be processed, and determining a normalization ratio; using the normalization ratio as the target noise standard deviation, and determining a feature point response value corresponding to the target noise standard deviation as a feature point screening threshold of the image to be processed; and using the feature point screening threshold to screen the feature points of the image to be processed.

[0006] In the second aspect, an embodiment of the present disclosure provides an image feature point screening device, including: an acquisition unit, used to acquire the image to be processed and the noise parameters of the image to be processed; a transformation unit, used to perform a variance stabilization transformation on the image to be processed according to the noise parameters, to obtain a transformed image, wherein the grayscale value range of the transformed image is different from the grayscale value range of the image to be processed; a first determination unit, used to normalize the grayscale value range of the transformed image to the grayscale value range of the image to be processed, and determine the normalization ratio; a second determination unit, used to use the normalization ratio as the target noise standard deviation, and determine the feature point response value corresponding to the target noise standard deviation as the feature point screening threshold of the image to be processed; a screening unit, used to use the feature point screening threshold to screen the feature points of the image to be processed.

[0007] In a third aspect, an embodiment of the present disclosure provides an electronic device comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the image feature point screening method as described in the first aspect.

[0008] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the image feature point screening method as described in the first aspect.

[0009] The image feature point screening method, device and electronic device provided by the embodiments of the present disclosure obtain the noise parameters of the image to be processed and the image to be processed; then, according to the noise parameters, the image to be processed is subjected to a variance stabilization transformation to obtain the transformed image; then, the grayscale value range of the transformed image is normalized to the grayscale value range of the image to be processed, and the normalization ratio is determined; then, the normalization ratio is used as the target noise standard deviation, and the feature point response value corresponding to the target noise standard deviation is determined as the feature point screening threshold of the image to be processed; finally, the feature points of the image to be processed are screened using the feature point screening threshold. In this way, the image is subjected to a noise variance stabilization transformation in combination with the noise parameters of the image, the feature point screening threshold corresponding to the image is calculated to distinguish between features and noise, and the extracted feature points are reasonably screened to retain high-quality feature points. In addition, extracting high-quality feature points helps to improve the efficiency, accuracy and stability of image matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that the originals and elements are not necessarily drawn to scale.

[0011] Figure 1 is a flowchart of an embodiment of the image feature point screening method according to the present disclosure;

[0012] Figure 2 is a flowchart of another embodiment of the image feature point screening method according to the present disclosure;

[0013] Figure 3 This is a flowchart of an embodiment of establishing a correspondence between noise standard deviation and feature point response value in the image feature point screening method disclosed herein;

[0014] Figure 4is a flowchart of another embodiment of the image feature point screening method according to the present disclosure;

[0015] Figure 5 is a structural diagram of an embodiment of an image feature point screening device according to the present disclosure;

[0016] Figure 6 is an exemplary system architecture diagram in which various embodiments of the present disclosure may be applied;

[0017] Figure 7 It is a structural diagram of a computer system suitable for implementing the electronic device of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0018] The following describes embodiments of the present disclosure in more detail with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of protection of the present disclosure.

[0019] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.

[0020] As used herein, the term "including" and its variations are open-ended, i.e., "including but not limited to." The term "based on" means "based, at least in part, on." The term "one embodiment" means "at least one embodiment," the term "another embodiment" means "at least one additional embodiment," and the term "some embodiments" means "at least some embodiments." Other terms are defined in the following description.

[0021] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0022] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, they should be understood as "one or more".

[0023] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0024] Please refer to Figure 1 , shows a process 100 of an embodiment of the image feature point screening method according to the present disclosure. The image feature point screening method includes the following steps:

[0025] Step 101: Obtain an image to be processed and noise parameters of the image to be processed.

[0026] In this embodiment, the execution entity of the image feature point screening method can obtain a target image and noise parameters of the target image. The target image is typically the image from which feature points are to be extracted. When a detector samples an image, it introduces noise into the image. Here, the noise parameters of the target image can be determined using a priori detector noise parameters and the image exposure gain.

[0027] Step 102: Perform a variance stabilization transformation on the image to be processed according to the noise parameters to obtain a transformed image.

[0028] In this embodiment, the execution entity may perform a variance-stabilizing transformation (VST) on the image to be processed based on the noise parameters to obtain a transformed image. Theoretically, the variance-stabilizing transformation can separate the variance of the estimator from the parameters. In practical applications, when the sample size is not particularly large, the variance-stabilizing transformation can improve the accuracy of the confidence interval to a certain extent.

[0029] As an example, the variance stabilization transform may be Generalize Anscombe Transform, which may be used to transform the image to be processed using the following formula (1):

[0030]

[0031] Wherein, Y represents the grayscale value of the image to be processed, f(Y) represents the grayscale value of the transformed image, k represents the Poisson noise parameter of the image to be processed, b represents the Gaussian noise parameter of the image to be processed, and g represents the black level of the image to be processed.

[0032] In this embodiment, the grayscale value range of the transformed image is usually different from the grayscale value range of the image to be processed. As an example, the grayscale value range of the image to be processed may be 0-255, and the grayscale value range of the transformed image may be 1-240.

[0033] Step 103 : normalize the grayscale value range of the transformed image to the grayscale value range of the image to be processed, and determine the normalization ratio.

[0034] In this embodiment, the execution entity may normalize the grayscale value range of the transformed image to the grayscale value range of the image to be processed, and determine a normalization ratio. Here, the normalization ratio is generally the ratio of the maximum value of the grayscale value range of the image to be processed to the maximum value of the grayscale value range of the image to be processed.

[0035] As an example, if the grayscale value range of the above-mentioned image to be processed is 0-255, and the grayscale value range of the above-mentioned transformed image is 1-240, the above-mentioned execution entity needs to normalize the grayscale value range 1-240 of the transformed image, that is, stretch the grayscale value range 1-240 to 0-255, and the corresponding normalization ratio is the ratio of 255 to 240.

[0036] Step 104 : Using the normalized magnification as the target noise standard deviation, determining a feature point response value corresponding to the target noise standard deviation as a feature point screening threshold for the image to be processed.

[0037] In this embodiment, the execution entity may use the normalization ratio as a target noise standard deviation, and determine a feature point response value corresponding to the target noise standard deviation as a feature point screening threshold of the image to be processed.

[0038] Here, the execution entity may store a correspondence between the noise standard deviation and the characteristic point response value, and the correspondence may be represented by a fitting relationship or a correspondence table. The execution entity may query the correspondence table for the characteristic point response value corresponding to the target noise standard deviation, or may substitute the target noise standard deviation into the fitting relationship to obtain the corresponding characteristic point response value.

[0039] Here, the purpose of performing a variance stabilization transformation on the image to be processed is to reduce the noise of the entire image to 1, that is, the noise standard deviation is normalized to 1. Normalizing the grayscale value range of the transformed image can be understood as scaling the grayscale value range, which is equivalent to scaling the noise standard deviation by the same amount. Therefore, the normalization factor can be used as the noise standard deviation.

[0040] Step 105 , using a feature point screening threshold, screens the feature points of the image to be processed.

[0041] In this embodiment, the execution entity may use the feature point screening threshold to screen the feature points of the image to be processed. Feature points whose response values are less than the feature point screening threshold may be deleted from the feature points of the image to be processed to obtain screened feature points, which may then be used for subsequent image processing.

[0042] Here, the execution entity may extract feature points from the image to be processed. For example, the feature point extraction may be performed based on a preset feature extraction algorithm. Feature extraction algorithms include, but are not limited to, FAST, Harris, SIFT (Scale-invariant feature transform), SURF (Speeded Up Robust Features), and other algorithms. Feature point response values may then be calculated for the extracted feature points. For example, the Harris corner detection algorithm may be used to calculate the feature point response values.

[0043] The method provided by the above embodiment of the present disclosure obtains the image to be processed and the noise parameters of the above image to be processed; then, according to the above noise parameters, the above image to be processed is subjected to variance stabilization transformation to obtain the transformed image; then, the grayscale value range of the above transformed image is normalized to the grayscale value range of the above image to be processed, and the normalization ratio is determined; then, the above normalization ratio is used as the target noise standard deviation, and the feature point response value corresponding to the above target noise standard deviation is determined as the feature point screening threshold of the above image to be processed; finally, the feature points of the above image to be processed are screened using the above feature point screening threshold. In this way, the noise variance stabilization transformation is performed on the image in combination with the noise parameters of the image, and the feature point screening threshold corresponding to the image is calculated to distinguish between features and noise, and the extracted feature points are reasonably screened to retain high-quality feature points. In addition, extracting high-quality feature points helps to improve the efficiency, accuracy and stability of image matching.

[0044] In some optional implementations, the noise introduced by detector sampling can be a single noise, such as Gaussian noise. In other embodiments, the noise introduced by detector sampling can be a combination of noise, such as Gaussian noise and Poisson noise. Gaussian noise is noise with a Gaussian distribution that is independent of pixel brightness, while Poisson noise is noise with a Poisson distribution that is dependent on pixel brightness. The noise parameters of the image to be processed can include Gaussian noise parameters and Poisson noise parameters.

[0045] In some optional implementations, the noise parameters of the above-mentioned image to be processed can be determined in the following manner: determining the Gaussian noise parameters of the above-mentioned image to be processed based on Gaussian noise prior parameters and the exposure gain of the above-mentioned image to be processed; determining the Poisson noise parameters of the above-mentioned image to be processed based on Poisson noise prior parameters and the exposure gain of the above-mentioned image to be processed.

[0046] Poisson noise, also known as shot noise, is a type of noise that can be modeled by a Poisson process. In electronics, Poisson noise is caused by the properties of discrete charges. Poisson noise also arises when optical devices count photons, due to the uncertainty in the arrival times of the photons.

[0047] The fundamental reason for Poisson noise is that light is composed of discrete photons (the particle nature of light). Photons emitted by a light source are received by a CMOS (Complementary Metal Oxide Semiconductor). A photoelectric conversion circuit converts the number of photons arriving per unit time into a voltage signal, which is then converted into a digital signal for output by an analog-to-digital conversion circuit. The more photons emitted by the light source per second that reach the CMOS, the larger the digital signal (or grayscale value) read out from that pixel. However, due to the uncertainty of photon arrival times, the detection results for the same light intensity signal still exhibit specific fluctuations, known as shot noise. For example, when the light intensity is low, say, 5 photons per second, the actual number of photons received by the CMOS per second may range from 0 to 10 (following a Poisson distribution).

[0048] Gaussian noise refers to a type of noise whose probability density function follows a Gaussian distribution (i.e., normal distribution). Common examples of Gaussian noise include readout noise, reset noise, and thermal noise.

[0049] The primary source of Gaussian noise in digital images arises from sensor (i.e., detector) noise during acquisition due to detection circuit imperfections and / or high temperatures. Since these various noise sources are diverse and statistically independent, the central limit theorem states that the spatial superposition of a large amount of this random noise will follow a Gaussian distribution. Due to the broadband nature of this noise, spatial low-pass filtering can be used in digital image processing to reduce Gaussian noise. However, low-pass filtering can blur fine-scale image edges and details because the low-pass filter also removes this high-frequency spatial information.

[0050] The Poisson noise prior parameters and Gaussian noise prior parameters can be understood as information that can be obtained through relevant technical means (mature technologies), such as Poisson modeling and Gaussian modeling. Image exposure gain is also information that can be obtained through relevant technical means, and the specific acquisition method is not limited here.

[0051] For example, the Poisson noise parameter k can be expressed as: k = kA*G + kB, and the Gaussian noise parameter b can be expressed as: b = bA*G^2 + bB*G + bC. Here, kA and kB are Poisson noise prior parameters, bA, bB, and bC are Gaussian noise prior parameters, and G is the image exposure gain.

[0052] In some optional implementations, the feature point response value may be a Harris corner response value. Here, image feature point extraction may be implemented based on a preset feature extraction algorithm. Feature extraction algorithms include, but are not limited to, FAST, Harris, SIFT, SURF, and other algorithms.

[0053] Furthermore, the feature point response values corresponding to the extracted multiple feature points may be response values corresponding to the corresponding feature extraction algorithms. In other embodiments, regardless of the feature extraction algorithm used, the feature point response values use Harris corner point response values.

[0054] As an optional implementation, the feature point response value is a Harris corner response value. Compared with other response values, the Harris corner response value can better reflect the robustness and quality of the feature points, thereby improving the quality of the feature points finally screened.

[0055] Continue to refer Figure 2 , which shows a process 200 of another embodiment of an image feature point screening method. The process 200 of the image feature point screening method includes the following steps:

[0056] Step 201: Obtain an image to be processed and noise parameters of the image to be processed.

[0057] Step 202: Perform a variance stabilization transformation on the image to be processed according to the noise parameters to obtain a transformed image.

[0058] Step 203 : normalize the grayscale value range of the transformed image to the grayscale value range of the image to be processed, and determine the normalization ratio.

[0059] In step 204 , the normalized magnification is used as the target noise standard deviation, and a feature point response value corresponding to the target noise standard deviation is determined as a feature point screening threshold of the image to be processed.

[0060] In this embodiment, steps 201-204 may be performed in a manner similar to steps 101-104, and are not described in detail here.

[0061] Step 205 : Deleting the feature points whose response values are less than the feature point screening threshold from the feature points of the image to be processed, to obtain a set of candidate feature points.

[0062] In this embodiment, the execution subject of the image feature point screening method may delete the feature points whose feature point response values are less than the feature point screening threshold value from the feature points of the image to be processed, to obtain a set of candidate feature points.

[0063] Here, the execution entity may extract feature points from the image to be processed. For example, image feature point extraction may be performed based on a preset feature extraction algorithm. Feature extraction algorithms include, but are not limited to, FAST, Harris, SIFT, SURF, and other algorithms. Feature point response values may then be calculated for the extracted feature points. For example, the Harris corner detection algorithm may be used to calculate the feature point response values.

[0064] Step 206 : sorting the feature points in the candidate feature point set in descending order of feature point response values.

[0065] In this embodiment, the execution entity may sort the feature points in the candidate feature point set in descending order of feature point response values.

[0066] Step 207 : performing spatial thinning processing on the feature points in the candidate feature point set according to the sorting result to obtain filtered feature points.

[0067] In this embodiment, the execution entity may perform spatial thinning on the feature points in the candidate feature point set according to the sorting result to obtain filtered feature points. As an example, the execution entity may perform spatial thinning on the feature points using a template. Specifically, an N*N (e.g., 5*5 or 10*10, etc.) interval box may be drawn with a certain feature point as the center to determine whether there are other feature points within the interval box. If so, the other feature points are deleted from the candidate feature point set until all feature points in the candidate feature point set are traversed, and the remaining feature points in the candidate feature point set are used as filtered feature points.

[0068] If the distance between feature points is close, the calculated descriptors may have a high similarity, which makes it very easy to search for mismatched feature points in the image to be matched; in addition, excessive concentration of feature points will also lead to fewer feature points distributed in other parts of the entire image, resulting in greater deviation in the solution of the overall pose of the entire image.

[0069] from Figure 2 It can be seen that Figure 1 Compared with the corresponding embodiment, the process 200 of the image feature point screening method in this embodiment embodies the step of performing spatial sparse processing on the feature points. Therefore, the solution described in this embodiment can prevent the feature points from being concentrated in a certain area in space and retain feature points with uniform spatial distribution.

[0070] Further references Figure 3, which shows a process 300 of an embodiment of establishing a correspondence between a noise standard deviation and a feature point response value in an image feature point screening method. The process 300 of establishing a correspondence between a noise standard deviation and a feature point response value includes the following steps:

[0071] Step 301: Acquire a target image, add noise with different noise standard deviations to the target image, and generate different noise maps.

[0072] In this embodiment, the entity that establishes the corresponding relationship can obtain a target image and add noise with different noise standard deviations to the target image to generate different noise maps. If M noises with different noise standard deviations are added, M noise maps will be generated, i.e., one noise map will be generated for each noise standard deviation.

[0073] Here, the brightness of all pixels of the target image is uniform. For example, the brightness of all pixels of the target image is 128.

[0074] It should be noted that the execution subject of the above-mentioned establishment of the corresponding relationship may be the execution subject of the above-mentioned image feature point screening method, or may be other electronic devices except the execution subject of the image feature point screening method.

[0075] Step 302 : determining the characteristic point response value of each noise map in different noise maps, and establishing a corresponding relationship between the added noise standard deviation and the characteristic point response value of the corresponding noise map.

[0076] In this embodiment, the execution entity may determine a characteristic point response value of each of the different noise maps, and establish a corresponding relationship between the added noise standard deviation and the characteristic point response value of the corresponding noise map.

[0077] As an example, feature points at the same position may be selected from different noise maps, and for each noise map, the feature point response value of the feature point at the selected position may be determined as the feature point response value of the noise map.

[0078] The method provided by the above-mentioned embodiments of the present disclosure obtains a target image, adds noise with different noise standard deviations to the target image, and generates different noise maps. The method then determines the feature point response value of each of the different noise maps and establishes a correspondence between the added noise standard deviation and the feature point response value of the corresponding noise map. This method establishes a correspondence between the noise standard deviation and the feature point response value, facilitating the subsequent use of this correspondence to determine a feature point screening threshold.

[0079] In some optional implementations, the execution entity may determine the feature point response value of each noise map in the different noise maps in the following manner: for each noise map in the different noise maps, the execution entity may extract multiple feature points from the noise map. Feature extraction algorithms include but are not limited to: FAST, Harris, SIFT, SURF and other algorithms. Afterwards, the feature point response value of each feature point may be determined. For example, the feature point response value may be calculated using the Harris corner detection algorithm. Then, the feature point response value of the noise map may be determined based on the feature point response values of the multiple feature points. As an example, feature points at the same position may be selected from different noise maps, and for the noise map, the feature point response value of the feature point at the selected position may be determined as the feature point response value of the noise map.

[0080] In some optional implementations, the execution entity may determine the feature point response value of the noise map based on the feature point response values of the multiple feature points in the following manner: the execution entity may determine the mean of the feature point response values of the multiple feature points as the feature point response value of the noise map; or the execution entity may determine the maximum of the feature point response values of the multiple feature points as the feature point response value of the noise map; or the execution entity may determine the standard deviation of the feature point response values of the multiple feature points as the feature point response value of the noise map. By selecting the mean, maximum, or standard deviation of the feature point response values of multiple feature points, the feature point response value of the noise map can be determined more accurately and flexibly.

[0081] In some optional implementations, the execution subject may determine the feature point response value of each of the different noise maps, and establish a corresponding relationship between the added noise standard deviation and the feature point response value of the corresponding noise map in the following manner: for each of the different noise maps, the execution subject may determine the image pyramid of the noise map. An image pyramid is a type of multi-scale image expression, and is an effective but conceptually simple structure for interpreting images at multiple resolutions. The image pyramid of an image is a set of image resolutions that are gradually reduced in a pyramid shape (from bottom to top) and originate from the same original image. It is obtained by stepwise downsampling until a certain termination condition is reached. We compare the layers of images to a pyramid. The higher the level, the smaller the image and the lower the resolution.

[0082] Next, the feature point response values for each layer of the image pyramid can be determined, thereby establishing a corresponding relationship between the noise standard deviation and the feature point response values for each layer. For example, if the image pyramid includes five layers, five corresponding relationships will be established, with one corresponding relationship between the noise standard deviation and the feature point response value for each layer.

[0083] In this way, the corresponding relationship between the noise standard deviation and the feature point response value corresponding to each layer of the image in the image pyramid can be established. When the feature points of the image to be processed are screened, the image to be processed can be processed by pyramid processing, and the noise standard deviation of each layer can be substituted into the corresponding relationship of the layer to determine the feature point screening threshold of the image layer, so as to perform feature point screening on the image layer. In this way, the feature point screening threshold of each layer of the image can be refined, so that the screened feature points are more reasonable.

[0084] Continue to see Figure 4 , Figure 4 This is a process of another embodiment of the image feature point screening method according to this embodiment. This process can represent an optional process when the image feature point screening method is applied in a specific application scenario. The process is summarized as follows:

[0085] Before screening the image feature points, the Harris response values under different degrees of noise are calibrated according to the pure noise images with different standard deviations, and a mapping relationship of "noise standard deviation-Harris response value" is constructed. The mapping relationship can be represented as a mapping table or a mapping relationship formula.

[0086] Input the current image, use the image exposure gain and detector noise parameters to determine the noise parameters of the current image; use the noise parameters of the current image to perform a noise variance stabilization transformation on the current image, and normalize the grayscale value range of the transformed image to the range of 0-255; substitute the normalization factor in the normalization process into the constructed mapping relationship, and calculate the noise Harris response threshold; extract the feature points of the current image, and calculate the Harris response value of the feature points; use the noise Harris response threshold to screen the feature points; sort the feature points after preliminary screening, perform spatial sparse processing on the feature points after preliminary screening according to the sorting result, and output the screened feature points.

[0087] Further references Figure 5 As an implementation of the methods shown in the above figures, the present application provides an embodiment of an image feature point screening device. Figure 1 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0088] like Figure 5As shown, the image feature point screening device 500 of this embodiment includes: an acquisition unit 501, a transformation unit 502, a first determination unit 503, a second determination unit 504, and a screening unit 505. The acquisition unit 501 is configured to acquire noise parameters of an image to be processed and the image to be processed; the transformation unit 502 is configured to perform a variance stabilization transformation on the image to be processed based on the noise parameters to obtain a transformed image, wherein the grayscale value range of the transformed image is different from the grayscale value range of the image to be processed; the first determination unit 503 is configured to normalize the grayscale value range of the transformed image to the grayscale value range of the image to be processed and determine a normalization factor; the second determination unit 504 is configured to use the normalization factor as a target noise standard deviation and determine a feature point response value corresponding to the target noise standard deviation as a feature point screening threshold for the image to be processed; and the screening unit 505 is configured to screen feature points of the image to be processed using the feature point screening threshold.

[0089] In this embodiment, the specific processing of the acquisition unit 501, the transformation unit 502, the first determination unit 503, the second determination unit 504 and the screening unit 505 of the image feature point screening device 500 can be referred to. Figure 1 These correspond to step 101, step 102, step 103, step 104 and step 105 in the embodiment.

[0090] In some optional implementations, the screening unit 505 can use the feature point screening threshold to screen the feature points of the image to be processed in the following manner: delete the feature points whose feature point response values are less than the feature point screening threshold in the feature points of the image to be processed to obtain a set of candidate feature points; sort the feature points in the set of candidate feature points in descending order of the feature point response values; and perform spatial sparse processing on the feature points in the set of candidate feature points according to the sorting result to obtain the screened feature points.

[0091] In some optional implementations, the correspondence between the noise standard deviation and the feature point response value can be established in the following manner: obtaining a target image, adding noise with different noise standard deviations to the above target image, generating different noise maps, wherein the brightness of all pixels of the above target image is uniform; determining the feature point response value of each noise map in the above different noise maps, and establishing a correspondence between the added noise standard deviation and the feature point response value of the corresponding noise map.

[0092] In some optional implementations, the feature point response value of each noise map in the above-mentioned different noise maps can be determined in the following manner: for each noise map in the above-mentioned different noise maps, multiple feature points are extracted from the noise map, the feature point response value of each feature point is determined, and the feature point response value of the noise map is determined based on the feature point response values of the above-mentioned multiple feature points.

[0093] In some optional implementations, the feature point response values of the noise map can be determined based on the feature point response values of the above-mentioned multiple feature points in the following manner: the mean of the feature point response values of the above-mentioned multiple feature points is determined as the feature point response value of the noise map; or the maximum value of the feature point response values of the above-mentioned multiple feature points is determined as the feature point response value of the noise map; or the standard deviation of the feature point response values of the above-mentioned multiple feature points is determined as the feature point response value of the noise map.

[0094] In some optional implementations, the feature point response value of each noise map in the above-mentioned different noise maps can be determined in the following manner, and a correspondence between the added noise standard deviation and the feature point response value of the corresponding noise map can be established: for each noise map in the above-mentioned different noise maps, an image pyramid of the noise map is determined, the feature point response value of each layer of the image in the above-mentioned image pyramid is determined, and a correspondence between the noise standard deviation corresponding to each layer of the image and the feature point response value is established.

[0095] In some optional implementations, the noise parameters of the image to be processed include Gaussian noise parameters and Poisson noise parameters; and the noise parameters of the image to be processed can be determined in the following manner: determining the Gaussian noise parameters of the image to be processed based on Gaussian noise prior parameters and the exposure gain of the image to be processed; determining the Poisson noise parameters of the image to be processed based on Poisson noise prior parameters and the exposure gain of the image to be processed.

[0096] In some optional implementations, the feature point response value is a Harris corner point response value.

[0097] See further Figure 6 , Figure 6 An exemplary system architecture 600 is shown to which an embodiment of the image feature point screening method disclosed herein can be applied.

[0098] like Figure 6 As shown, system architecture 600 may include terminal devices 6011, 6012, and 6013, a network 602, and a server 603. Network 602 is used to provide a medium for communication links between terminal devices 6011, 6012, and 6013 and server 603. Network 602 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0099] Users can use terminal devices 6011, 6012, and 6013 to interact with server 603 via network 602 to send or receive messages, etc. For example, server 603 can receive images to be processed sent by terminal devices 6011, 6012, and 6013, and terminal devices 6011, 6012, and 6013 can also obtain the corresponding relationship between the noise standard deviation and the feature point response value from server 603. Various communication client applications can be installed on terminal devices 6011, 6012, and 6013, such as image capture applications, video processing applications, video playback applications, instant messaging software, etc.

[0100] Terminal devices 6011, 6012, and 6013 can be hardware or software. When terminal devices 6011, 6012, and 6013 are hardware, they can be various electronic devices with cameras and supporting information interaction, including but not limited to smartphones, tablet computers, laptop computers, etc. When terminal devices 6011, 6012, and 6013 are software, they can be installed in the electronic devices listed above. They can be implemented as multiple software or software modules (for example, multiple software or software modules for providing distributed services), or they can be implemented as a single software or software module. No specific limitation is made here.

[0101] Terminal devices 6011, 6012, and 6013 can obtain the noise parameters of the image to be processed and the above-mentioned image to be processed; then, based on the above-mentioned noise parameters, the above-mentioned image to be processed is subjected to variance stabilization transformation to obtain the transformed image; then, the grayscale value range of the above-mentioned transformed image is normalized to the grayscale value range of the above-mentioned image to be processed, and the normalization ratio is determined; then, the above-mentioned normalization ratio is used as the target noise standard deviation, and the feature point response value corresponding to the above-mentioned target noise standard deviation is determined as the feature point screening threshold of the above-mentioned image to be processed; finally, the feature points of the above-mentioned image to be processed are screened using the above-mentioned feature point screening threshold.

[0102] Server 603 can be a server that provides various services. For example, it can be a backend server that screens feature points of an image to be processed. Server 603 can obtain the image to be processed and the noise parameters of the image to be processed from terminal devices 6011, 6012, and 6013. Then, based on the noise parameters, the server 603 performs a variance stabilization transformation on the image to be processed to obtain a transformed image. Then, the grayscale value range of the transformed image is normalized to the grayscale value range of the image to be processed, and a normalization factor is determined. Then, the normalization factor is used as the target noise standard deviation, and the feature point response value corresponding to the target noise standard deviation is determined as the feature point screening threshold of the image to be processed. Finally, the feature point screening threshold is used to screen the feature points of the image to be processed.

[0103] It should be noted that the server 603 can be hardware or software. When the server 603 is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or it can be implemented as a single server. When the server 603 is software, it can be implemented as multiple software or software modules (for example, for providing distributed services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0104] It should also be noted that the image feature point screening method provided in the embodiment of the present disclosure can be executed by the terminal devices 6011, 6012, and 6013. In this case, the image feature point screening device is usually set in the terminal devices 6011, 6012, and 6013. The image feature point screening method provided in the embodiment of the present disclosure can also be executed by the server 603. In this case, the image feature point screening device is usually set in the server 603.

[0105] It should be understood that Figure 6 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0106] Reference below Figure 7 , which shows an electronic device (eg, Figure 6 The terminal device in the embodiments of the present disclosure may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (such as in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 7 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0107] like Figure 7 As shown, the electronic device 700 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 701, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 702 or a program loaded from a storage device 708 into a random access memory (RAM) 703. Various programs and data required for the operation of the electronic device 700 are also stored in the RAM 703. The processing device 701, the ROM 702, and the RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0108] Typically, the following devices may be connected to the I / O interface 705: an input device 706 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 707 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 708 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 709. The communication device 709 may allow the electronic device 700 to communicate with other devices wirelessly or by wire to exchange data. Although Figure 7 The electronic device 700 is shown with various devices, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead. Figure 7 Each block shown in the figure may represent one device, or may represent multiple devices as needed.

[0109] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 709, or installed from the storage device 708, or installed from the ROM 702. When the computer program is executed by the processing device 701, the above-mentioned functions defined in the method of the embodiment of the present disclosure are executed. It should be noted that the computer-readable medium described in the embodiment of the present disclosure can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection having one or more conductors, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In embodiments of the present disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In embodiments of the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including, but not limited to, an electromagnetic signal, an optical signal, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wire, optical cable, RF (radio frequency), etc., or any suitable combination thereof.

[0110] The computer-readable medium may be included in the electronic device, or may exist independently and not be incorporated into the electronic device. The computer-readable medium carries one or more programs. When executed by the electronic device, the electronic device: obtains an image to be processed and noise parameters of the image to be processed; performs a variance stabilization transform on the image to be processed based on the noise parameters to obtain a transformed image, wherein the grayscale value range of the transformed image is different from the grayscale value range of the image to be processed; normalizes the grayscale value range of the transformed image to the grayscale value range of the image to be processed and determines a normalization factor; uses the normalization factor as a target noise standard deviation and determines a feature point response value corresponding to the target noise standard deviation as a feature point screening threshold for the image to be processed; and uses the feature point screening threshold to screen feature points of the image to be processed.

[0111] Computer program code for performing the operations of embodiments of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0112] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0113] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware. The units described may also be provided in a processor. For example, they may be described as follows: a processor includes an acquisition unit, a transformation unit, a first determination unit, a second determination unit, and a screening unit. The names of these units do not, in some cases, constitute limitations on the units themselves. For example, the acquisition unit may also be described as "acquiring an image to be processed and noise parameters of the image to be processed."

[0114] The above description is merely a preferred embodiment of the present disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by a specific combination of the above-mentioned technical features, but should also encompass other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned inventive concept. For example, a technical solution formed by mutually replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present disclosure.

Claims

1. A method for screening image feature points, characterized in that: include: Acquiring an image to be processed and noise parameters of the image to be processed; performing a variance stabilization transform on the image to be processed according to the noise parameter to obtain a transformed image, wherein a grayscale value range of the transformed image is different from a grayscale value range of the image to be processed; Normalizing the grayscale value range of the transformed image to the grayscale value range of the image to be processed, and determining a normalization ratio; Taking the normalized magnification as a target noise standard deviation, determining a feature point response value corresponding to the target noise standard deviation as a feature point screening threshold of the image to be processed; The feature points of the image to be processed are screened using the feature point screening threshold.

2. The method according to claim 1, characterized in that The step of screening the feature points of the image to be processed by using the feature point screening threshold comprises: Deleting feature points whose feature point response values are less than the feature point screening threshold from the feature points of the image to be processed, to obtain a set of candidate feature points; Sorting the feature points in the candidate feature point set in descending order of feature point response values; According to the sorting result, spatial sparse processing is performed on the feature points in the candidate feature point set to obtain filtered feature points.

3. The method according to claim 1, characterized in that The corresponding relationship between the noise standard deviation and the characteristic point response value is established as follows: Acquire a target image, add noise with different noise standard deviations to the target image, and generate different noise maps, wherein the brightness of all pixels of the target image is uniform; Determine the characteristic point response value of each noise map in the different noise maps, and establish a corresponding relationship between the added noise standard deviation and the characteristic point response value of the corresponding noise map.

4. The method according to claim 3, characterized in that Determining the characteristic point response value of each noise map in the different noise maps includes: For each noise map in the different noise maps, multiple feature points are extracted from the noise map, a feature point response value of each feature point is determined, and a feature point response value of the noise map is determined based on the feature point response values of the multiple feature points.

5. The method according to claim 4, characterized in that The determining, based on the feature point response values of the plurality of feature points, the feature point response values of the noise map includes: Determine the average of the feature point response values of the plurality of feature points as the feature point response value of the noise image; or Determine the maximum value of the feature point response values of the plurality of feature points as the feature point response value of the noise image; or The standard deviation of the feature point response values of the plurality of feature points is determined as the feature point response value of the noise image.

6. The method according to claim 3, characterized in that The determining of the characteristic point response value of each noise map in the different noise maps and establishing a corresponding relationship between the added noise standard deviation and the characteristic point response value of the corresponding noise map includes: For each noise map in the different noise maps, an image pyramid of the noise map is determined, a feature point response value of each layer of the image in the image pyramid is determined, and a corresponding relationship between a noise standard deviation corresponding to each layer of the image and the feature point response value is established.

7. The method according to claim 1, characterized in that The noise parameters of the image to be processed include Gaussian noise parameters and Poisson noise parameters; and The noise parameters of the image to be processed are determined in the following manner: Determining Gaussian noise parameters of the image to be processed according to a Gaussian noise priori parameter and the exposure gain of the image to be processed; The Poisson noise parameters of the image to be processed are determined according to the Poisson noise priori parameters and the exposure gain of the image to be processed.

8. The method according to any one of claims 1 to 7, characterized in that The feature point response value is the Harris corner point response value.

9. An image feature point screening device, characterized in that: include: An acquisition unit, configured to acquire an image to be processed and noise parameters of the image to be processed; a transform unit, configured to perform a variance stabilization transform on the image to be processed according to the noise parameter to obtain a transformed image, wherein a grayscale value range of the transformed image is different from a grayscale value range of the image to be processed; a first determining unit, configured to normalize the grayscale value range of the transformed image to the grayscale value range of the image to be processed, and determine a normalization ratio; a second determining unit, configured to use the normalization magnification as a target noise standard deviation, and determine a feature point response value corresponding to the target noise standard deviation as a feature point screening threshold of the image to be processed; The screening unit is used to screen the feature points of the image to be processed by using the feature point screening threshold.

10. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 8.

11. A computer-readable medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 8 is implemented.

Images