Image feature extraction method and device, medium and electronic equipment
The Gaussian filtering function performs noise processing and data point uniformization processing on the image, which solves the problem of uneven distribution of image feature extraction in the intelligent driving system and improves the positioning accuracy.
Patent Information
- Application Number
- CN202411167460.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-23
- Publication Date
- 2024-12-03
AI Technical Summary
In intelligent driving and autonomous driving systems, the image feature points perceived by the sensor generate noise due to interference factors such as light intensity, resulting in inaccurate feature extraction and uneven distribution of feature points, which affects the positioning accuracy.
The initial image is noise-processed by using Gaussian filtering function, and the standard deviation of the filtering function is dynamically determined to reduce noise sensitivity, and the data point distribution uniformization process is performed on the denoised image to improve the accuracy of feature extraction.
It effectively reduces the noise influence in image feature extraction, improves the uniformity of feature points distribution, and improves the positioning accuracy of intelligent driving and autonomous driving systems.
Smart Images

Figure CN119068434A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of vehicle-mounted image processing, and in particular to an image feature extraction method, device, medium and electronic equipment. Background Art
[0002] With the continuous development of intelligent and autonomous driving technologies, more and more vehicles are equipped with intelligent driving devices or autonomous driving systems. In order to ensure the normal operation of intelligent driving devices and autonomous driving systems, it is necessary to install sensors on the vehicle to perceive the surrounding environmental data and information during driving. For example, perception cameras are used to observe road image data to calculate the vehicle's location information in the map and plan the vehicle's driving path.
[0003] However, the environmental data perceived by the sensor will cause a lot of noise in the image feature points (such as image edges, corners, textures, etc.) due to interference factors such as light intensity, which affects the accuracy of feature processing and extraction. In addition, the existing image feature extraction will have the problem of large-scale clustering and overlap of extracted feature points in some areas, but fewer or even none in other areas, resulting in uneven overall distribution of feature points, which in turn causes a decrease in the accuracy of subsequent positioning and intelligent driving.
[0004] Therefore, there is an urgent need for a method that can effectively improve the accuracy of feature extraction from image data. Summary of the invention
[0005] In order to solve the above technical problems, the present application is proposed. The embodiments of the present application provide an image feature extraction method, device, medium and electronic device.
[0006] According to one aspect of the present application, there is provided an image feature extraction method, comprising: acquiring an initial image; wherein the initial image comprises a plurality of data points within a specific area; performing noise processing on the initial image using a Gaussian filter function to obtain a denoised image; wherein the standard deviation of the Gaussian filter function is dynamically determined based on the pixel variance of all data points within a first preset range of the corresponding data points; performing data point distribution uniformity processing on the denoised image to obtain a target image.
[0007] In one embodiment, the use of a Gaussian filter function to perform noise processing on the initial image to obtain a denoised image includes: calculating the pixel variance of all data points within a first preset range of the single data point; calculating the standard deviation of the Gaussian filter function corresponding to the single data point based on the pixel variance of all data points within the first preset range of the single data point; wherein the standard deviation of the Gaussian filter function corresponding to the single data point is inversely correlated with the pixel variance of all data points within the first preset range of the single data point; calculating the weight values of all data points within a second preset range of the single data point based on the standard deviation of the Gaussian filter function corresponding to the single data point; wherein the second preset range is included in the first preset range; calculating the denoised value of the single data point based on the weight values of all data points within the second preset range of the single data point; wherein the denoised value represents the pixel value after noise processing is performed on the single data point.
[0008] In one embodiment, the denoised value of the single data point is calculated based on the weight values of all data points within the second preset range of the single data point, including: based on the weight values of all data points within the second preset range of the single data point, weighted summing up the pixel values of all data points within the second preset range of the single data point to obtain the denoised value of the single data point.
[0009] In one embodiment, the method of calculating the standard deviation of the Gaussian filter function corresponding to the single data point based on the pixel variance of all data points within the first preset range of the single data point includes: calculating the standard deviation of the Gaussian filter function corresponding to the single data point based on the pixel variance of all data points within the first preset range of the single data point and the corresponding Gaussian filter function; wherein the contrast value between the pixel variance of all data points within the first preset range of the single data point and the corresponding Gaussian filter function remains unchanged.
[0010] In one embodiment, the calculation of the weight values of all data points within the second preset range of the single data point based on the standard deviation of the Gaussian filter function corresponding to the single data point includes: calculating the weight values of all data points within the second preset range of the single data point based on the standard deviation of the Gaussian filter function corresponding to the single data point and the corresponding Gaussian filter function.
[0011] In one embodiment, the process of performing data point distribution uniformization processing on the denoised image to obtain the target image includes: based on the distribution concentration of the data points in the denoised image, deleting some data points whose distribution concentration is greater than a preset concentration threshold to obtain the target image.
[0012] In one embodiment, based on the distribution concentration of data points in the denoised image, some data points whose distribution concentration is greater than a preset concentration threshold are deleted to obtain the target image, including: using a window of preset size to frame data points in the denoised image to obtain multiple window areas; if the number of data points in a single window area is greater than a preset number threshold, calculating the interval distance between any two data points in the window area; if there are two data points whose interval distance is less than the preset distance threshold, deleting the data point with the smaller pixel value among the two data points to obtain the target image.
[0013] According to another aspect of the present application, an image feature extraction device is provided, comprising: an image acquisition module, used to acquire an initial image; wherein the initial image includes multiple data points in a specific area; a noise processing module, used to perform noise processing on the initial image using a Gaussian filter kernel to obtain a denoised image; wherein the standard deviation of each of the data points is determined based on the variance of all data points within a preset range of the data point; and a data uniformity module, used to perform data point distribution uniformity processing on the denoised image to obtain a target image.
[0014] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and the computer program is used to execute any of the above-mentioned image feature extraction methods.
[0015] According to another aspect of the present application, an electronic device is provided, comprising: a processor; a memory for storing instructions executable by the processor; and the processor is used to execute any of the above-mentioned image feature extraction methods.
[0016] The present application provides an image feature extraction method, device, medium and electronic device, which obtain an initial image; wherein the initial image includes multiple data points in a specific area; use a Gaussian filter function to perform noise processing on the initial image to obtain a denoised image; wherein the standard deviation of the Gaussian filter function is dynamically determined based on the pixel variance of all data points within a first preset range of the corresponding data points; perform data point distribution uniformity processing on the denoised image to obtain a target image; that is, after the initial image is obtained, use a Gaussian filter function to perform noise processing on the initial image to filter out noise, and dynamically determine the standard deviation of the Gaussian filter function based on the pixel variance of data points within the preset range of data points, so as to effectively reduce the sensitivity of the data points to noise according to the degree of discreteness of the data points, thereby improving the accuracy of image feature extraction; at the same time, perform data distribution uniformity processing on the denoised image to improve the problem of uniform distribution of image feature points, thereby further improving the accuracy of image feature extraction. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0018] Figure 1 It is a flowchart of an image feature extraction method provided by an exemplary embodiment of the present application.
[0019] Figure 2 It is a structural schematic diagram of data point uniformization processing provided by an exemplary embodiment of the present application.
[0020] Figure 3 It is a structural schematic diagram of an image feature extraction device provided by an exemplary embodiment of the present application.
[0021] Figure 4 It is a structural diagram of an electronic device provided by an exemplary embodiment of the present application. DETAILED DESCRIPTION
[0022] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.
[0023] Figure 1 FIG. 1 is a flow chart of an image feature extraction method provided by an exemplary embodiment of the present application. Figure 1 As shown, the image feature extraction method includes the following steps:
[0024] Step 110: Acquire an initial image.
[0025] Among them, the initial image includes multiple data points in a specific area. Specifically, the present application can use the on-board camera to perceive the surrounding environment in real time to construct the initial image. However, there will be a large amount of random white noise interference in the environmental image data perceived by the on-board camera. If the noise is not processed, it will significantly affect the accuracy of the vehicle's perception of the surrounding environment, resulting in poor intelligent driving or automatic driving effects. Therefore, after the on-board camera obtains the initial image, the present application processes the initial image to extract the feature data points in the initial image, thereby providing accurate and concise environmental image data for subsequent intelligent driving and automatic driving.
[0026] Step 120: Use a Gaussian filter function to perform noise processing on the initial image to obtain a denoised image.
[0027] The standard deviation of the Gaussian filter function is dynamically determined based on the pixel variance of all data points within a first preset range of the corresponding data points.
[0028] Gaussian filtering is a linear smoothing filtering technology, which is widely used in the noise reduction process of image processing. It works by weighted averaging each pixel in the image and the pixels in its neighborhood, where the weight is calculated according to the Gaussian function. The weight distribution of Gaussian filtering is affected by the standard deviation. The larger the standard deviation, the wider the weight distribution, the stronger the filtering effect, and the better the noise can be eliminated; the smaller the standard deviation, the more concentrated the weight distribution, the relatively weaker the smoothing effect on the image, and the better the image details can be preserved. The selection of standard deviation is crucial to the effect of Gaussian filtering. A larger standard deviation will lead to a wider weight distribution, thereby performing a stronger smoothing process on the image; while a smaller standard deviation will produce a more localized weight distribution, which has less impact on the image and helps to retain more detail information. Therefore, in practical applications, it is necessary to select a suitable standard deviation according to the specific image processing requirements. However, the standard deviation of the Gaussian filter function is usually pre-set. Whether the standard deviation is set appropriately directly affects the effect of Gaussian filtering, and Gaussian filter functions with different standard deviations have different processing effects on different noise interferences.
[0029] Since the degree of discreteness of the data points in the image area is large, the corresponding pixel variance is also large. At this time, in order to retain the detail information such as the differences of the data points in the image area, the standard deviation of the Gaussian filter function should be appropriately reduced; when the degree of discreteness of the data points in the image area is small, the corresponding pixel variance is also small. At this time, in order to improve the smoothness of the data points in the image area, the standard deviation of the Gaussian filter function should be appropriately increased. The present application dynamically adjusts the standard deviation of the Gaussian filter function to dynamically determine the standard deviation of the Gaussian filter function according to the pixel variance of all data points within a first preset range of the data points, so as to adopt different standard deviations for different data points, thereby reducing the sensitivity of the data points to noise, and removing areas where feature points such as prominent edges and corners that are more irrelevant to details in the initial image are concentrated, thereby improving the quality and stability of the image feature points, and providing accurate basic data for subsequent vehicles to build maps of the road environment.
[0030] Step 130: Perform data point distribution uniformization processing on the denoised image to obtain a target image.
[0031] After using the Gaussian filter function to perform noise processing on the initial image, the present application further performs data point distribution uniformity processing on the denoised image to minimize the impact of excessive differences in the degree of aggregation of data points between different image regions on the accuracy of overall feature extraction.
[0032] The present application provides an image feature extraction method, which is obtained by acquiring an initial image; wherein the initial image includes multiple data points in a specific area; using a Gaussian filter function to perform noise processing on the initial image to obtain a denoised image; wherein the standard deviation of the Gaussian filter function is dynamically determined based on the pixel variance of all data points within a first preset range of the corresponding data points; performing data point distribution uniformization processing on the denoised image to obtain a target image; that is, after the initial image is acquired, using a Gaussian filter function to perform noise processing on the initial image to filter out noise, and dynamically determining the standard deviation of the Gaussian filter function based on the pixel variance of the data points within the preset range of the data points, so as to effectively reduce the sensitivity of the data points to noise according to the degree of discreteness of the data points, thereby improving the accuracy of image feature extraction; and at the same time, performing data distribution uniformization processing on the denoised image to improve the problem of uniform distribution of image feature points, thereby further improving the accuracy of image feature extraction.
[0033] In one embodiment, the specific implementation method of the above-mentioned step 120 may be: calculating the pixel variance of all data points within a first preset range of a single data point; based on the pixel variance of all data points within the first preset range of the single data point, calculating the standard deviation of the Gaussian filter function corresponding to the single data point; wherein the standard deviation of the Gaussian filter function corresponding to the single data point is inversely correlated with the pixel variance of all data points within the first preset range of the single data point; based on the standard deviation of the Gaussian filter function corresponding to the single data point, calculating the weight values of all data points within a second preset range of the single data point; wherein the second preset range is included in the first preset range; based on the weight values of all data points within the second preset range of the single data point, calculating the denoised value of the single data point; wherein the denoised value represents the pixel value after noise processing of the single data point.
[0034] For a single data point, the present application can calculate the pixel variance of all data points (including the single data point) within the first preset range of the single data point based on the pixel values of all data points within the first preset range of the single data point. Specifically, taking the calculation method of the pixel variance within the (2k+1)×(2k+1) neighborhood of the single data point (i, j) as an example, the calculation formula of the pixel variance D(i, j) within the (2k+1)×(2k+1) neighborhood of the data point (i, j) is as follows:
[0035]
[0036] Among them, I(i+m,j+n) is the value of the data point in the mth row and nth column in the (2k+1)×(2k+1) neighborhood of the single data point (i,j), and μ(i,j) is the mean of the pixel values of all data points in the (2k+1)×(2k+1) neighborhood of the single data point (i,j).
[0037] After calculating the pixel variance of all data points within the first preset range of the single data point, the standard deviation of the Gaussian filter function corresponding to the single data point is calculated based on the pixel variance of all data points within the first preset range of the single data point. Optionally, the present application may use the following formula to calculate the standard deviation σ(i, j) of the Gaussian filter function corresponding to the single data point:
[0038]
[0039] Among them, σ base is the preset basic standard deviation value, and α is the coefficient that controls the adjustment range (α>0).
[0040] After calculating the standard deviation of the Gaussian filter function corresponding to the single data point, the weight values of all data points within the second preset range of the single data point are further calculated based on the standard deviation of the Gaussian filter function corresponding to the single data point, and the second preset range is included in the first preset range. It should be understood that the first preset range and the second preset range in the present application can be set according to actual needs. For example, the first preset range and the second preset range are the same, for example, the (2k+1)×(2k+1) neighborhood of the above data point (i, j) is 3×3 (corresponding to k=1) or 5×5 (corresponding to k=2).
[0041] After calculating the weight values of all data points within the second preset range of the single data point, the denoising value of the single data point is calculated based on the weight values of all data points within the second preset range of the single data point, thereby implementing a denoising operation on the single data point.
[0042] In one embodiment, the specific calculation method of the denoising value of a single data point can be: based on the weight values of all data points within the second preset range of the single data point, weighted sum of the pixel values of all data points within the second preset range of the single data point to obtain the denoising value of the single data point.
[0043] After calculating the weight values of all data points within the second preset range of the single data point, the weighted average of multiple data points within the second preset range of the single data point is used as the pixel value of the single data point to reduce the impact of noise interference on the single data point.
[0044] In one embodiment, the specific calculation method of the standard deviation of the Gaussian filter function corresponding to a single data point can be: based on the pixel variance of all data points within a first preset range of the single data point and the corresponding Gaussian filter function, the standard deviation of the Gaussian filter function corresponding to the single data point is calculated; wherein the contrast value between the pixel variance of all data points within the first preset range of the single data point and the corresponding Gaussian filter function remains unchanged.
[0045] After calculating the pixel variance of all data points within the first preset range of the single data point, the pixel variance is compared with the traditional two-dimensional Gaussian filter function P(i, j; σ) in real time to obtain a contrast function F(i, j), wherein the contrast function can be calculated using the following formula:
[0046]
[0047] Since α, σ base and D(i,j) are constants when the data point (i,j) is determined, then the contrast function F(i,j) is a function of the Gaussian convolution kernel k and the standard deviation σ. If F(i,j) = 1, it means that the degree of discreteness of the data points in the neighborhood of the single data point is the same as the degree of discreteness of the two-dimensional Gaussian filter function. In this application, F(i,j) is set to a fixed value (for example, set to 1). The larger the pixel variance D(i,j), the smaller the standard deviation σ. The above formula can be used to calculate the standard deviation of the corresponding Gaussian filter function when F(i,j) is fixed, that is, a unique standard deviation of the Gaussian filter function can be calculated for each data point, so as to adaptively and dynamically adjust the standard deviation of the Gaussian filter function, so as to better filter out the noise in the initial image and improve the accuracy of the image data.
[0048] In one embodiment, the specific calculation method of the weight values of all data points within the second preset range of a single data point can be: based on the standard deviation of the Gaussian filter function corresponding to the single data point and the corresponding Gaussian filter function, calculate the weight values of all data points within the second preset range of the single data point.
[0049] After calculating the standard deviation of the Gaussian filter function corresponding to the single data point, the standard deviation of the Gaussian filter function is substituted into the Gaussian filter function, and the weight values of all data points within the second preset range of the single data point are calculated based on the Gaussian filter function, and then the denoising value of the single data point is calculated based on the weight values of all data points within the second preset range of the single data point.
[0050] In one embodiment, the specific implementation of the above step 130 may be: based on the distribution concentration of data points in the denoised image, some data points whose distribution concentration is greater than a preset concentration threshold are deleted to obtain the target image.
[0051] In the process of image feature extraction, there may be overlapping data points in some areas (i.e., the distribution of data points in this area is relatively large), while there are fewer or even no data points in other areas (i.e., the distribution of data points in this area is relatively small), which leads to inaccurate feature extraction. Therefore, this application deletes some data points whose distribution concentration is greater than the preset concentration threshold to improve the problem of overlapping data points as much as possible, thereby further improving the accuracy of image feature extraction.
[0052] In one embodiment, the specific implementation method of the above step 130 may be: use a window of preset size to select data points in the denoised image to obtain multiple window areas; if the number of data points in a single window area is greater than a preset number threshold, calculate the interval distance between any two data points in the window area; if there are two data points with an interval distance less than the preset distance threshold, delete the data point with the smaller pixel value between the two data points to obtain the target image.
[0053] Specifically, Figure 2 As shown, the present application can set a window of a preset size (e.g., 3*3, 5*5, etc.) to select data points on each frame of the denoised image to obtain multiple window areas (e.g., Figure 2 The dotted box shown in the figure) is used, and the distribution and aggregation degree of the data points in the single window area is tested, that is, whether the number of data points in the single window area is greater than the preset number threshold. If it is greater, it means that the number of data points in the single window area is large (that is, there may be cluster overlap). Then, the interval distance between any two data points in the window area is further calculated to obtain the interval distance between all the data points in the window area. If there are two data points whose interval distance is less than the preset distance threshold (for example, 1.5 times the radius of the filter box of the non-maximum suppression algorithm), the data points with smaller pixel values (such as Figure 2 The hollow data points shown are retained to retain data points with larger pixel values, thereby minimizing the aggregation and overlap of data points and improving the uniformity of data point distribution.
[0054] Figure 3 FIG. 1 is a schematic diagram of the structure of an image feature extraction device provided by an exemplary embodiment of the present application. Figure 3 As shown, the image feature extraction device 30 includes: an image acquisition module 31, used to acquire an initial image; wherein the initial image includes multiple data points in a specific area; a noise processing module 32, used to perform noise processing on the initial image using a Gaussian filter kernel to obtain a denoised image; wherein the standard deviation of each data point is determined based on the variance of all data points within a preset range of the data point; a data uniformity module 33, used to perform data point distribution uniformity processing on the denoised image to obtain a target image.
[0055] The present application provides an image feature extraction device, which acquires an initial image through an image acquisition module 31; wherein the initial image includes multiple data points in a specific area; the noise processing module 32 uses a Gaussian filter function to perform noise processing on the initial image to obtain a denoised image; wherein the standard deviation of the Gaussian filter function is dynamically determined based on the pixel variance of all data points within a first preset range of the corresponding data points; the data uniformity module 33 performs data point distribution uniformity processing on the denoised image to obtain a target image; that is, after the initial image is acquired, the Gaussian filter function is used to perform noise processing on the initial image to filter out noise, and the standard deviation of the Gaussian filter function is dynamically determined according to the pixel variance of the data points within the preset range of the data points, so as to effectively reduce the sensitivity of the data points to noise according to the degree of discreteness of the data points, thereby improving the accuracy of image feature extraction; at the same time, the denoised image is subjected to data distribution uniformity processing to improve the problem of uniform distribution of image feature points, thereby further improving the accuracy of image feature extraction.
[0056] In one embodiment, the above-mentioned noise processing module 32 can be further configured as: calculating the pixel variance of all data points within a first preset range of a single data point; based on the pixel variance of all data points within the first preset range of the single data point, calculating the standard deviation of the Gaussian filter function corresponding to the single data point; wherein the standard deviation of the Gaussian filter function corresponding to the single data point is inversely correlated with the pixel variance of all data points within the first preset range of the single data point; based on the standard deviation of the Gaussian filter function corresponding to the single data point, calculating the weight values of all data points within a second preset range of the single data point; wherein the second preset range is included in the first preset range; based on the weight values of all data points within the second preset range of the single data point, calculating the denoised value of the single data point; wherein the denoised value represents the pixel value after noise processing of the single data point.
[0057] In one embodiment, the noise processing module 32 may be further configured to weighted sum the pixel values of all data points within the second preset range of the single data point based on the weight values of all data points within the second preset range of the single data point to obtain a denoised value of the single data point.
[0058] In one embodiment, the noise processing module 32 can be further configured as follows: based on the pixel variance of all data points within the first preset range of the single data point and the corresponding Gaussian filter function, the standard deviation of the Gaussian filter function corresponding to the single data point is calculated; wherein the contrast value between the pixel variance of all data points within the first preset range of the single data point and the corresponding Gaussian filter function remains unchanged.
[0059] In one embodiment, the noise processing module 32 may be further configured to calculate weight values of all data points within a second preset range of a single data point based on the standard deviation of the Gaussian filter function corresponding to the single data point and the corresponding Gaussian filter function.
[0060] In one embodiment, the data uniformity module 33 may be further configured to: based on the distribution concentration of data points in the denoised image, delete some data points whose distribution concentration is greater than a preset concentration threshold to obtain a target image.
[0061] In one embodiment, the above-mentioned data uniformity module 33 can be further configured as follows: use a window of preset size to select data points in the denoised image to obtain multiple window areas; if the number of data points in a single window area is greater than a preset number threshold, then calculate the distance between any two data points in the window area; if there are two data points with a distance less than the preset distance threshold, then delete the data point with the smaller pixel value between the two data points to obtain the target image.
[0062] Below, reference Figure 4 The electronic device according to the embodiment of the present application is described. The electronic device may be any one or both of the first device and the second device, or a stand-alone device independent of them, and the stand-alone device may communicate with the first device and the second device to receive the collected input signal from them.
[0063] Figure 4 A block diagram of an electronic device according to an embodiment of the present application is illustrated.
[0064] like Figure 4 As shown, the electronic device 10 includes one or more processors 11 and a memory 12 .
[0065] The processor 11 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
[0066] The memory 12 may include one or more computer program products, and the computer program product may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, a random access memory (RAM) and / or a cache memory (cache), etc. The non-volatile memory may include, for example, a read-only memory (ROM), a hard disk, a flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 11 may run the program instructions to implement the methods of the various embodiments of the present application described above and / or other desired functions. Various contents such as input signals, signal components, noise components, etc. may also be stored in the computer-readable storage medium.
[0067] In one example, the electronic device 10 may further include: an input device 13 and an output device 14, and these components are interconnected via a bus system and / or other forms of connection mechanisms (not shown).
[0068] When the electronic device is a stand-alone device, the input device 13 may be a communication network connector, which is used to receive the collected input signals from the first device and the second device.
[0069] In addition, the input device 13 may also include, for example, a keyboard, a mouse, and the like.
[0070] The output device 14 can output various information to the outside, including the determined distance information, direction information, etc. The output device 14 can include, for example, a display, a speaker, a printer, a communication network and a remote output device connected thereto, and the like.
[0071] Of course, to simplify, Figure 4 Only some of the components related to the present application in the electronic device 10 are shown, and components such as a bus, an input / output interface, etc. are omitted. In addition, according to specific application situations, the electronic device 10 may also include any other appropriate components.
[0072] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps of the method according to various embodiments of the present application described in the above-mentioned "Exemplary Method" section of this specification.
[0073] The computer program product may be written in any combination of one or more programming languages to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages, such as Java, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as an independent software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.
[0074] In addition, an embodiment of the present application may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the steps of the method according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.
[0075] The computer readable storage medium can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.
[0076] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.
[0077] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.
[0078] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.
[0079] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
[0080] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.
Claims
1. A method for extracting image features, characterized in that: include: Acquire an initial image; wherein the initial image includes a plurality of data points within a specific area; Using a Gaussian filter function to perform noise processing on the initial image to obtain a denoised image; wherein the standard deviation of the Gaussian filter function is dynamically determined based on the pixel variance of all data points within a first preset range of the corresponding data points; The denoised image is processed for data point distribution uniformity to obtain a target image.
2. The image feature extraction method according to claim 1, characterized in that: The adopting of a Gaussian filter function to perform noise processing on the initial image to obtain a denoised image comprises: Calculate the pixel variance of all data points within a first preset range of a single data point; Based on the pixel variance of all data points within the first preset range of the single data point, the standard deviation of the Gaussian filter function corresponding to the single data point is calculated; wherein the standard deviation of the Gaussian filter function corresponding to the single data point is inversely correlated with the pixel variance of all data points within the first preset range of the single data point; Calculating weight values of all data points within a second preset range of the single data point based on the standard deviation of the Gaussian filter function corresponding to the single data point; wherein the second preset range is included in the first preset range; Based on the weight values of all data points within the second preset range of the single data point, a denoised value of the single data point is calculated; wherein the denoised value represents a pixel value after noise processing is performed on the single data point.
3. The image feature extraction method according to claim 2, characterized in that: The calculating the denoised value of the single data point based on the weight values of all data points within the second preset range of the single data point comprises: Based on the weight values of all data points within the second preset range of the single data point, the pixel values of all data points within the second preset range of the single data point are weighted summed to obtain a denoised value of the single data point.
4. The image feature extraction method according to claim 2, characterized in that: The calculating the standard deviation of the Gaussian filter function corresponding to the single data point based on the pixel variance of all data points within the first preset range of the single data point comprises: Based on the pixel variance of all data points within the first preset range of the single data point and the corresponding Gaussian filter function, the standard deviation of the Gaussian filter function corresponding to the single data point is calculated; wherein the contrast value between the pixel variance of all data points within the first preset range of the single data point and the corresponding Gaussian filter function remains unchanged.
5. The image feature extraction method according to claim 2, characterized in that: The calculating the weight values of all data points within a second preset range of the single data point based on the standard deviation of the Gaussian filter function corresponding to the single data point comprises: Based on the standard deviation of the Gaussian filter function corresponding to the single data point and the corresponding Gaussian filter function, the weight values of all data points within the second preset range of the single data point are calculated.
6. The image feature extraction method according to claim 1, characterized in that: The performing data point distribution uniformization processing on the denoised image to obtain a target image comprises: Based on the distribution concentration of data points in the denoised image, some data points whose distribution concentration is greater than a preset concentration threshold are deleted to obtain the target image.
7. The image feature extraction method according to claim 6, characterized in that: The step of deleting some data points whose distribution concentration is greater than a preset concentration threshold based on the distribution concentration of data points in the denoised image to obtain the target image comprises: Using a window of a preset size to select data points in the denoised image to obtain multiple window areas; If the number of data points in a single window area is greater than a preset number threshold, then calculating the interval distance between any two data points in the window area; If there are two data points whose interval distance is less than a preset distance threshold, the data point with the smaller pixel value among the two data points is deleted to obtain the target image.
8. An image feature extraction device, characterized in that: include: An image acquisition module, used to acquire an initial image; wherein the initial image includes a plurality of data points within a specific area; A noise processing module, used for performing noise processing on the initial image using a Gaussian filter kernel to obtain a denoised image; wherein the standard deviation of each of the data points is determined based on the variance of all data points within a preset range of the data point; The data uniformity module is used to delete some data points whose distribution concentration is greater than a preset concentration threshold based on the distribution concentration of the data points in the denoised image to obtain a target image.
9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and the computer program is used to execute the image feature extraction method described in any one of claims 1 to 7.
10. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is used to execute the image feature extraction method described in any one of claims 1 to 7.
