WO2019223069A1 - 基于直方图的虹膜图像增强方法、装置、设备及存储介质 - Google Patents
基于直方图的虹膜图像增强方法、装置、设备及存储介质 Download PDFInfo
- Publication number
- WO2019223069A1 WO2019223069A1 PCT/CN2018/094397 CN2018094397W WO2019223069A1 WO 2019223069 A1 WO2019223069 A1 WO 2019223069A1 CN 2018094397 W CN2018094397 W CN 2018094397W WO 2019223069 A1 WO2019223069 A1 WO 2019223069A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- iris image
- iris
- pixel
- gray value
- enhanced
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 230000003044 adaptive effect Effects 0.000 claims abstract description 28
- 238000012545 processing Methods 0.000 claims description 36
- 239000011159 matrix material Substances 0.000 claims description 25
- 230000001186 cumulative effect Effects 0.000 claims description 21
- 238000005315 distribution function Methods 0.000 claims description 19
- 230000001965 increasing effect Effects 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000002708 enhancing effect Effects 0.000 claims description 8
- 238000001514 detection method Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 16
- 230000000694 effects Effects 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 11
- 238000012549 training Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 7
- 230000009466 transformation Effects 0.000 description 7
- 238000009826 distribution Methods 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/18—Eye characteristics, e.g. of the iris
- G06V40/193—Preprocessing; Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
- G06T5/94—Dynamic range modification of images or parts thereof based on local image properties, e.g. for local contrast enhancement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Definitions
- the present application relates to the field of image processing, and in particular, to a histogram-based iris image enhancement method, device, device, and storage medium.
- iris has the characteristics of uniqueness, stability, collectability and non-invasiveness.
- high-definition iris images are often used as training sets, but due to the limitations of acquisition equipment and changes in the acquisition environment, the quality of the acquired iris images will be poor, such as low contrast and noise interference. Such issues will affect the highlighting of iris texture features, and then affect the clarity and recognition efficiency of the iris image training set.
- the overall contrast of the collected iris image is usually adjusted dynamically.
- the accuracy of the iris image processed in the recognition system is still not high.
- a histogram-based iris image enhancement method includes:
- the iris image set includes an iris image, and the iris image includes a user identifier
- the first enhanced iris image in the first enhanced iris image set is sharpened by using a Laplacian to obtain a second enhanced iris image set.
- a histogram-based iris image enhancement device includes:
- An iris image set acquisition module configured to obtain an iris image set, wherein the iris image set includes an iris image, and the iris image includes a user identifier;
- the iris sequence acquisition module is used to calculate the contrast of the iris images in the iris image set, and sort the iris images corresponding to each user ID in the iris image set in the order of the contrast, to obtain the corresponding Initial iris sequence;
- the initial iris set acquisition module is used to obtain a preset number of iris images from the initial iris sequence corresponding to each user ID in order of increasing contrast, to form an initial iris set;
- a first enhanced iris image set acquisition module configured to use an adaptive histogram equalization algorithm to locally enhance the initial iris image in the initial iris set to obtain a first enhanced iris image set;
- a second enhanced iris image set acquisition module is configured to sharpen the first enhanced iris image in the first enhanced iris image set by using a Laplacian to obtain a second enhanced iris image set.
- a computer device includes a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor.
- the processor executes the computer-readable instructions, the following steps are implemented:
- the iris image set includes an iris image, and the iris image includes a user identifier
- the first enhanced iris image in the first enhanced iris image set is sharpened by using a Laplacian to obtain a second enhanced iris image set.
- One or more non-volatile readable storage media storing computer-readable instructions, which when executed by one or more processors, cause the one or more processors to perform the following steps:
- the iris image set includes an iris image, and the iris image includes a user identifier
- the first enhanced iris image in the first enhanced iris image set is sharpened by using a Laplacian to obtain a second enhanced iris image set.
- FIG. 1 is an application scenario diagram of an iris image enhancement method based on a histogram in an embodiment of the present application
- FIG. 2 is a flowchart of a method for enhancing an iris image based on a histogram in an embodiment of the present application
- step S10 in FIG. 2 is a flowchart of a specific implementation of step S10 in FIG. 2;
- step S20 in FIG. 2 is a flowchart of a specific implementation of step S20 in FIG. 2;
- step S50 in FIG. 2 is a flowchart of a specific implementation of step S50 in FIG. 2;
- FIG. 6 (a) is an example diagram of an initial iris image in the embodiment of the present application.
- FIG. 6 (b) is an example diagram of a second enhanced iris image in the embodiment of the present application.
- FIG. 7 is a schematic diagram of an iris image enhancement device based on a histogram in an embodiment of the present application.
- FIG. 8 is a schematic diagram of a computer device in an embodiment of the present application.
- the iris image enhancement method based on the histogram provided in this application can be applied in a computer device or system to enhance the iris image to solve the problem of low accuracy of iris image recognition.
- the computer device may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices.
- the system may include a server and a client.
- FIG. 1 shows an application scene diagram of the histogram-based iris image enhancement method applied in a system.
- the server and the client are connected through the network, the client collects or obtains the iris image, and the server obtains the iris image from the client.
- the client can be a camera, camera, scanner, or other device with a photographing function. (Phone, tablet, etc.), or a storage device that stores iris images.
- the server can be implemented by a server or a server cluster composed of a plurality of servers.
- a histogram-based iris image enhancement method is provided.
- the method is applied to a computer device as an example for description, and includes the following steps:
- the iris image set includes an iris image, and the iris image includes a user logo.
- the iris image set refers to an image set composed of iris images, and the iris image refers to an image obtained by photographing an iris inside a user's eye through a camera device.
- the iris image set may be acquired in real time, or may be stored in a computer device in advance.
- An iris image set may include iris images of one user, and may also include iris images of multiple users.
- an iris image set includes iris images of N users. If each user has M iris images, then the iris image set has M * N iris images.
- the iris image set includes at least two iris images.
- the user logo refers to the logo of the user to which the iris image belongs, and is used to classify the iris image according to the user. Each iris image corresponds to a user logo, and the iris image of the same user corresponds to the same user logo.
- a predetermined number of iris images can be acquired by photographing the eyes of the user to form an iris image set, or a predetermined number of pre-stored iris images can be acquired from a computer device to form an iris image set, or acquired by photographing the eyes of the user Part of the iris image, and another part of the pre-stored iris image is obtained from the computer equipment, and the two together constitute the iris image set.
- multiple iris images of several users can be selected as the iris image set during the same time period. For example, it can be collected at noon on a rainy day or in the afternoon on a sunny day. This can avoid the problem that the contrast of the different iris images in the acquired iris image set is greatly different due to light changes.
- S20 Calculate the contrast of the iris images in the iris image set, and sort the iris images corresponding to each user ID in the iris image set in the descending order of the contrast to obtain the initial iris sequence corresponding to each user ID.
- contrast is an index to measure image quality.
- the contrast of an iris image is the ratio of black and white of the image, which is used to represent the gradation of the iris image from black to white.
- the larger the ratio the more the gradation of the iris image from black to white, and the richer the color expression.
- the effect of contrast on visual effects is very critical.
- the larger the contrast the clearer and sharper the image, and the brighter and more vivid the colors.
- the high-contrast iris image has more obvious advantages in detail performance, sharpness, and high-speed moving object performance in some dark scenes.
- the initial iris sequence refers to an iris sequence in which the iris image corresponding to each user's logo is arranged in order of increasing contrast.
- the iris sequence corresponding to the same user logo is arranged in descending order of contrast, which is the sub-iris sequence corresponding to the user logo.
- the contrast calculation is performed on the multiple iris images of each user identifier one by one, and the multiple iris images corresponding to each user identifier are sorted according to the order of the contrast from large to small, to obtain the sub-substance corresponding to each user identifier. Iris sequence.
- the larger the contrast the higher the sharpness of the iris image. Therefore, the iris images corresponding to each user's logo in the iris image set are arranged in order of increasing contrast.
- each user ID includes M iris images
- these N sub-iris sequences together form the initial iris sequence, and the iris in each sub-iris sequence
- the images are arranged in order of increasing contrast.
- iris image with higher contrast is used as a criterion for selecting the iris image, which is convenient for subsequent selection of the iris image with higher contrast for processing to obtain an iris image with more texture characteristics.
- S30 Obtain a preset number of iris images from the initial iris sequence corresponding to each user ID according to the order of increasing contrast, and form an initial iris set.
- the preset number is a preset value, which is used to select a certain number of iris images for subsequent enhancement processing.
- the preset number may be set according to a requirement of a training sample size. For example, if the number of training samples required for each user identification is P in subsequent model training, a preset number can be set to P.
- a preset number of values may be set according to the number of iris images corresponding to each user identifier. For example, in the case of 30 images collected by each user's logo, the corresponding preset number can be set to 10, that is, only the 30 iris images need to be displayed in the order of the contrast in each user logo.
- 10 iris images can be selected as the initial iris set from large to small.
- the initial iris set is a set composed of a preset number of iris images with a contrast value selected from the initial iris sequence.
- the iris image with larger contrast is selected to form the initial iris set, thereby excluding iris images with lower contrast, reducing some redundant images, reducing the workload of subsequent iris image enhancement processing, accelerating the speed of iris image enhancement processing, and improving subsequent iris.
- Image processing efficiency At the same time, because the iris image with a large contrast is selected as the initial iris set, the enhancement degree of the iris image can be improved, and the recognition rate of the iris image can be improved.
- S40 Use an adaptive histogram equalization algorithm to locally enhance the initial iris image in the initial iris set to obtain a first enhanced iris image set.
- the Adaptive Histogram Equalization (AHE) algorithm uses moving sub-window technology to perform histogram equalization on the sub-window area containing the pixels of the initial iris image, that is, the minimum gray value and Highest gray value, equalize the initial iris image histogram in the area corresponding to the sub-window, and according to the gray-level mapping principle of the entire initial iris image histogram equalization, give the image to be processed in the area corresponding to the sub-window Pixel reassignment.
- the gray value of the initial iris image is generally low, and its histogram is distributed in the area where the gray value is small. Therefore, it is relatively concentrated in the low gray area.
- the gray value is increased. Distribution range, and the distance between each gray level is also increased, more uniform throughout the gray level, effectively expanding the gray distribution range of the initial iris image, making the initial iris image brightness increase to a certain extent Improved the contrast of the initial iris image.
- Local enhancement refers to the process of enhancing the local information of the image.
- the histogram equalization algorithm is used to disperse all sub-block regions of the initial iris image.
- the local detail information of the iris image is adaptively enhanced through the superposition of the histogram equalization calculation of the sub-block area. Therefore, the adaptive histogram equalization algorithm is used to locally enhance the initial iris image, so that the initial iris image retains the image details as much as possible, and the local texture details can also be adaptively adjusted to enhance the internal details of the image and the entire image.
- the detailed information of the iris image is more clear, so that the first enhanced iris image set with rich and clear texture can be obtained, which can improve the subsequent recognition accuracy.
- the gray value of the initial iris image in the initial iris set is the input of the adaptive histogram equalization algorithm
- the gray value of the first enhanced iris image is the output of the adaptive histogram equalization algorithm.
- S50 The first enhanced iris image in the first enhanced iris image set is sharpened by using a Laplacian operator to obtain a second enhanced iris image set.
- the first enhanced iris image refers to an iris image obtained by using an adaptive histogram equalization algorithm to enhance the initial iris image in the initial iris set.
- Laplacian operator is a second-order differential operator, which is suitable for improving image blur caused by diffuse reflection of light.
- the principle is that during the process of shooting and recording an image, light spots diffusely reflect light to its surrounding area. This diffuse reflection of light causes a certain degree of blurring in the image.
- the degree of blurring is relative to that of images taken under normal circumstances. It is said that it is often a constant multiple of the Laplace operator. Therefore, sharpening the Laplace operator on the image can reduce the blur of the image and improve the sharpness of the image. Therefore, by sharpening the first enhanced iris image, highlighting the edge detail features of the first enhanced iris image, and improving the contour definition of the first enhanced iris image.
- Sharpening processing refers to the transformation of sharpening an image to enhance the target boundaries and image details in the image.
- the second enhanced iris image refers to an iris image obtained by performing a sharpening process on the iris image in the first enhanced iris concentration using a Laplacian operator. After the first enhanced iris image is sharpened by the Laplace operator, the image edge details are enhanced, and the sharpness of the bright area edge of the first enhanced iris image is also improved, thereby protecting the first enhanced iris image. detail.
- the process of sharpening the first enhanced iris image by using a Laplacian may be: using a Laplacian to obtain a second derivative of a gray value of a pixel of the first enhanced iris image, and a second order The pixel corresponding to the derivative at zero is the edge pixel of the image.
- Such processing can more clearly show the edges of the iris texture, so as to obtain a clear iris training set with richer texture details and improve the recognition effect.
- an iris image set is first obtained, a contrast calculation is performed on the iris image set in the iris image set, and an iris image with a high contrast in the iris image set is extracted to form an initial iris set, thereby reducing the iris image with poor quality and reducing
- the redundant operation is helpful to improve the enhancement degree of the iris image and the efficiency of subsequent enhancement processing.
- use the adaptive histogram equalization algorithm to locally enhance the initial iris image in the initial iris set, adjust the dynamic range of the gray value of the initial iris image, better highlight the local features, and refine the internal details.
- the details of the iris image generally improve the contrast of the iris image and make the texture features of the iris image clearer, so as to obtain a clear iris training set with richer texture details and improve the accuracy of subsequent recognition.
- step S10 obtaining an iris image set includes the following steps:
- S11 Obtain the measured distance between the human eye and the camera in real time. If the measured distance is not within the distance threshold, a prompt message is sent.
- the measured distance refers to the distance between the user's eyes and the camera
- the distance threshold refers to a preset distance value given by repeated tests during the experimental measurement. If the subject is at the distance threshold, the data collected by the computer equipment The image quality is better than the images acquired at other locations.
- the distance threshold range refers to the limit set above and below the distance threshold. It is easy to understand that the measured distance can also capture a clear image within a certain range where the distance threshold fluctuates. Under the condition that the image quality is clear, in order to facilitate the fast shooting of the image, a range of the distance threshold ⁇ a% can be set as the distance threshold range. Optionally, a can be 5, 10, 15, and so on.
- the prompt information is used to prompt the user to make corresponding adjustments in order to take a clear image.
- the prompt information includes, but is not limited to, arrow identification (such as arrows in different directions), text prompt information (such as too far, appropriate or close), and voice Information (such as "Please approach the camera", “Collecting” or "Please stay away from the camera”).
- the measured distance After obtaining the measured distance, determine whether the measured distance is within the distance threshold. If it is not within the distance threshold, send a prompt message, and the user adjusts accordingly according to the prompt information, and then obtain the measured distance until the measured distance is within the distance threshold. By guiding the user within the range of the distance threshold, it is beneficial to improve the quality of the subsequently acquired iris image.
- the focal length of the infrared camera can be used as the distance threshold.
- the infrared camera can be manually or automatically adjusted to determine the focal length of the camera according to the sharpness of the image. .
- the corresponding prompt information is sent by comparing the measured distance and the distance threshold range, which can guide the user to quickly adjust the position and improve the efficiency of iris image collection.
- the quality of the iris obtained by shooting is better. Continuous shooting can obtain multiple iris images of the same person, which is convenient and fast, and provides a better quality iris image set for subsequent enhancement processing.
- the measured distance between the user's eyes and the camera is obtained, and the measured distance is compared with the distance threshold range, and corresponding prompt information is fed back to the user according to the comparison result.
- the user adjusts according to the prompt information, and when the measured distance is within the threshold range
- the camera is controlled for continuous shooting to obtain the iris image, which can quickly and easily obtain the iris image, which also improves the quality of the iris image.
- step S20 the contrast of the iris image in the iris image set is calculated, and specifically includes the following steps:
- a pixel is a basic element of a digital image, and a pixel is obtained by discretizing a continuous space when an analog image is digitized.
- Each pixel has integer row (height) and integer column (width) position coordinates, while each pixel has integer grayscale or color values.
- An image is made up of many pixels.
- digital image data can be represented by a matrix, so matrix theory and matrix algorithms can be used to analyze and process digital images.
- the pixel information of a grayscale image is a matrix, the rows of the matrix correspond to the height of the image, the columns of the matrix correspond to the width of the image, and the matrix elements correspond to the pixels of the image.
- the value of the matrix element is the grayscale value of the pixel, which represents the grayscale image.
- a gray value corresponding to each pixel of the iris image can be obtained through an image information acquisition tool. That is, the path corresponding to the image is given, and the image under the path is read through the path. For example, this can be achieved with the imread function:
- jpg is the format of the image
- lean is the name of the image
- D: ⁇ is the path of the lean image
- I is the matrix corresponding to the lean image.
- the center pixel is the pixel located at the center in a given area. In this implementation, sequentially referring to each pixel as the center pixel means that in a given area, each pixel in the area is regarded as the center pixel. For example, if there are 15 pixels in the area, and these 15 pixels are used as the center pixels in turn, then there are 15 center pixels.
- the boundary pixel When the boundary pixel is the center pixel, the boundary pixel can be regarded as the center pixel by extending the pixel, that is, the gray value of a pixel that does not exist in the neighborhood of the boundary pixel is set to be equal to the gray value of the boundary pixel.
- the matrix of an iris image is:
- the gray value of the pixels in the first row and the first column is 22, and there are no pixels in the left and upper parts.
- the gray values of the left and upper pixels are set to the boundary.
- the gray value of the same pixel size, that is, the gray value of the left and upper parts are both 22.
- the neighborhood pixel refers to the pixel adjacent to the center pixel position.
- the pixel p at the coordinate (x, y) has two horizontal and two vertical adjacent pixels, and each pixel distance (x, y) One unit distance.
- the coordinates are: (x-1, y), (x + 1, y), (x, y-1), (x, y + 1).
- This pixel set is defined as the four neighborhoods of pixel p, which is represented by N4 (p).
- pixel p has 4 diagonally adjacent pixels with coordinates: (x-1, y-1), (x + 1, y-1), (x-1, y + 1), (x + 1, y + 1).
- These four diagonally adjacent pixels and N4 (p) are collectively referred to as the 8-neighborhood of pixel P and are represented by N8 (P).
- the difference between the gray value of each central pixel and the pixel of the corresponding neighborhood is 4. If the gray value of the central pixel is h (x, y) Display, then the difference between the gray value of the pixel and the corresponding pixel in the 4 neighborhoods can be obtained by the following formula:
- the number k of differences between gray values can be obtained by the following formula:
- the number k of differences in gray values can be obtained by the following formula:
- the contrast of the iris image is represented by C, and the differences in gray values are q 1 , q 2 ... q k , and k is a positive integer.
- the specific calculation formula for the contrast C of the iris image is as follows:
- the contrast C is a specific value.
- the gray value of each pixel of the iris image is obtained, and each pixel is sequentially used as the center pixel.
- the gray value of the center pixel and the gray value of the preset neighborhood pixel are calculated Degree difference
- the number of gray value differences is calculated by presetting the size of the neighborhood and the number of rows and columns of the corresponding matrix of the iris image, and then comparing the gray value of each central pixel in the iris image with The gray value difference corresponding to the neighboring pixels is squared and summed and then divided by the number of gray value difference values.
- the obtained result is the contrast of the iris image.
- steps S21 to S24 the contrast of the iris image can be calculated simply and quickly, and a high-quality iris image can be filtered out by comparing the contrast.
- an adaptive histogram equalization algorithm is used to locally enhance the initial iris image in the initial iris set to obtain a first enhanced iris image set, which specifically includes:
- Performing histogram equalization processing on corresponding pixel blocks in the initial iris image specifically includes:
- the pixel block is subjected to gray transform processing to obtain an equalized pixel block.
- the preset sub-tile refers to a preset mobile sub-window, which is used as a standard for adjusting the contrast of a pixel block. Understandably, the smaller the preset sub-tile, the smaller the amount of information it contains, and the greater the degree of contrast stretching of the pixel block. For example, if the size of the preset sub-tile A is 32 ⁇ 32 and the size of the preset sub-tile B is 8 ⁇ 8, then the contrast of the pixel block corresponding to the preset sub-tile B is stretched more than the preset sub-picture. The degree of stretching of the contrast of the pixel block corresponding to the block A.
- a pixel block is a sub-tile obtained by dividing an initial iris image.
- the initial iris image size is 64 ⁇ 64, and if it is divided equally by 2 ⁇ 2, there are 4 pixel blocks, and the size of each pixel block is 32 ⁇ 32.
- the preset sub-block position order refers to the order in which each pixel block of the sub-block is located, such as from top to bottom and then from left to right, or from left to right and then from top to bottom order of.
- an initial iris image is evenly divided according to M ⁇ N, and the order of the positions of the preset sub-tiles may be: the order of the first row, the second row ... the Mth row, or the first column, the second column ... Order of the Nth column.
- the preset sub-tile is moved on the pixel block of the initial iris image, and iterates in the order of the position of the pixel block to perform histogram equalization on the pixel block.
- the block is operated, and the local information of the iris image is adaptively adjusted, which increases the internal details of the initial iris image.
- the histogram of an image is a statistical description describing the frequency of occurrence of each gray value in the image, and it reflects the number of occurrences of each gray value in an image.
- the specific method of grayscale transformation processing is to use the calculated cumulative distribution function as a transformation function to perform grayscale transformation on each pixel block in the initial iris image through the cumulative distribution function, thereby transforming the relatively concentrated areas of grayscale values to relatively Large areas increase the contrast of the pixel blocks in the original iris image.
- a histogram of a pixel block can be obtained through an image information tool, and then the number of each gray value of the histogram corresponding to the pixel block is counted; a cumulative distribution function of each gray value is calculated; based on the cumulative distribution function Perform gray-scale transformation processing on the pixel block to obtain an equalized pixel block.
- the gray scale transformation is performed by the cumulative transformation function, and the histogram distribution of the pixel blocks of the initial iris image is changed to a uniform distribution histogram distribution.
- the gray value is adjusted by using the cumulative function to achieve the contrast enhancement effect of the initial iris image. Understandably, the histogram equalization of the corresponding pixel block in the initial iris image is achieved through these three steps.
- a preset sub-block is traversed in accordance with the position order of the sub-blocks on the pixel blocks of the initial iris image, and then a histogram equalization process is performed so that the dynamic range of the pixel gray value of the initial iris image is obtained Efficient expansion, and local information is also adaptively adjusted, enhancing internal details.
- step S50 the first enhanced iris image in the first enhanced iris image set is sharpened by using a Laplacian operator, which specifically includes the following steps:
- S51 Obtain the gray value of each pixel of the first enhanced iris image in the first enhanced iris image set, and use the Laplacian to sharpen the gray value of each pixel to obtain the sharpened pixels. grayscale value.
- the first enhanced iris image in the first enhanced iris image set can be directly read to obtain the gray value of each iris image pixel.
- the specific reading method is similar to step S21, and is not repeated here.
- the Laplace operator based on second-order differential is defined as:
- Each pixel gray value of the gray value R (x, y) of the first enhanced iris image is sharpened according to the following formula to obtain the sharpened pixel gray value.
- S52 Obtain a corresponding second enhanced iris image based on the sharpened pixel gray value in the first enhanced iris image.
- the sharpened pixel gray value is replaced with the gray value at the original (x, y) pixel to obtain a second enhanced iris image.
- the Laplace operator Four-neighbor sharpening template matrix Laplace operator sharpening is performed on a first enhanced iris image in the first enhanced iris image set using a four-neighbor sharpening template matrix H.
- the original iris image (Fig. 6 (a)) is enhanced by adaptive histogram equalization algorithm enhancement and Laplace operator sharpening.
- the iris image is a comparison view of the second enhanced iris image (Fig. 6 (b)). It can be seen that the overall contrast of the initial iris image is low. Compared with the initial iris image, the dynamic range of the gray value of the second enhanced iris image is effectively adjusted, and the unclear local information in 6 (a) is better presented. The local detail information is prominent (the iris image in the human eye image is visually clear), and the edge details are relatively rich.
- the gray value of each pixel of the first enhanced iris image in the first enhanced iris image set is obtained, and Laplacian sharpening processing is performed to obtain the sharpened pixel gray value.
- a corresponding second enhanced iris image is obtained.
- the iris image that has been enhanced by the adaptive histogram equalization algorithm is sharpened with Laplacian.
- the edge features of the image are enhanced while suppressing the noise introduced during the enhancement of the first enhanced iris image, thereby protecting it. Details of the first enhanced iris image.
- the above steps are not only simple and convenient, and improve the real-time performance of the iris image processing, but also the edge features of the second enhanced iris image are more prominent after processing, the overall contrast of the iris image set is greatly improved, and the texture characteristics of the iris image are enhanced. , It is helpful to improve the accuracy of iris image recognition.
- the recognition algorithm is implemented by calculating the Euclidean distance or by a Support Vector Machine (SVM) classifier.
- SVM Support Vector Machine
- the comparison recognition rate is calculated as the histogram-based Enhancement effect of iris image enhancement algorithm.
- the results show that the recognition rate of the unprocessed iris image is 83%, and the recognition rate of the iris image processed by the histogram-based iris image enhancement method in this embodiment is 99.3%, and the recognition rate is increased by 16.3%.
- a histogram-based iris image enhancement device corresponds to the histogram-based iris image enhancement method in the embodiment described above.
- the histogram-based iris image enhancement device includes an iris image set acquisition module 10, an iris sequence acquisition module 20, an initial iris set acquisition module 30, a first enhanced iris image set acquisition module 40, and a second enhanced iris Image set acquisition module 50.
- the implementation functions of the iris image set acquisition module 10, the iris sequence acquisition module 20, the initial iris set acquisition module 30, the first enhanced iris image set acquisition module 40, and the second enhanced iris image set acquisition module 50 are based on the above-mentioned embodiment.
- the steps corresponding to the iris image enhancement method of the histogram correspond one by one. In order to avoid redundant description, this embodiment is not detailed one by one.
- the iris image set acquisition module 10 is configured to acquire an iris image set, where the iris image set includes an iris image, and the iris image includes a user identifier.
- the iris sequence acquisition module 20 is used to calculate the contrast of the iris images in the iris image set, and sort the iris images corresponding to each user ID in the iris image set in the order of the contrast, to obtain the initial corresponding to each user ID. Iris sequence.
- the initial iris set acquisition module 30 is configured to obtain a preset number of iris images from the initial iris sequence corresponding to each user identifier according to the order of increasing contrast, to form an initial iris set.
- a first enhanced iris image set acquisition module 40 is configured to perform local enhancement processing on an initial iris image in an initial iris set using an adaptive histogram equalization algorithm to obtain a first enhanced iris image set.
- a second enhanced iris image set acquisition module 50 is configured to sharpen the first enhanced iris image in the first enhanced iris image set using a Laplacian operator to obtain a second enhanced iris image set.
- the iris image set acquisition module 10 includes a measured distance detection unit 11 and an iris image set acquisition unit 12.
- the measured distance detection unit 11 is configured to obtain the measured distance of the human eye and the camera in real time, and if the measured distance is not within the distance threshold, a prompt message is sent.
- the iris image set obtaining unit 12 is configured to control the camera to continuously shoot if the actual measured distance is within a distance threshold, to obtain an iris image set.
- the iris sequence acquisition module 20 further includes a contrast calculation unit 21 for calculating the contrast of the iris image in the iris image set.
- the contrast calculation unit 21 includes a gray value acquisition sub-unit 211, a gray value difference acquisition sub unit 212, a gray number difference number acquisition sub unit 213, and a contrast calculation sub unit 214.
- the gray value acquisition subunit 211 is configured to acquire a gray value of each pixel of the iris image in the iris image set, and sequentially use each pixel as a central pixel.
- the gray value difference obtaining subunit 212 is configured to calculate a difference between a gray value of each center pixel and a gray value of a corresponding neighbor pixel according to a preset neighborhood size.
- the number-of-gray-values acquisition subunit 213 is configured to obtain the number of differences between the gray-scale values in the iris image based on the preset neighborhood size and the number of rows and columns of the corresponding matrix of the iris image.
- the contrast calculation subunit 214 is configured to perform a square sum of the difference between the gray value of each central pixel in the iris image and the gray value of the corresponding neighboring pixel, and divide it by the number of differences Number to obtain the contrast of the iris image.
- the first enhanced iris image set obtaining module 40 is configured to obtain preset sub-tiles, and perform histogram equalization processing on corresponding pixel blocks in the initial iris image according to the position order of the preset sub-tiles;
- Performing histogram equalization processing on corresponding pixel blocks in the initial iris image specifically includes:
- the pixel block is subjected to gray transform processing to obtain an equalized pixel block.
- the second enhanced iris image set acquisition module 50 includes a sharpened gray value acquisition unit 51 and a second enhanced iris image acquisition unit 52.
- the sharpened gray value obtaining unit 51 is configured to obtain the gray value of each pixel of the first enhanced iris image in the first enhanced iris image set, and use the Laplacian to determine the gray value of each pixel. Perform sharpening to obtain the gray value of the sharpened pixel.
- the second enhanced iris image acquisition unit 52 is configured to acquire a corresponding second enhanced iris image based on the sharpened pixel gray value in the first enhanced iris image.
- Each module in the above histogram-based iris image enhancement device may be implemented in whole or in part by software, hardware, and a combination thereof.
- the above-mentioned modules may be embedded in the hardware in or independent of the processor in the computer device, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.
- a computer device is provided.
- the computer device may be a client, and its internal structure diagram may be as shown in FIG. 8.
- the computer device includes a processor, a memory, and a network interface connected through a system bus.
- the processor of the computer device is used to provide computing and control capabilities.
- the memory of the computer device includes a non-volatile storage medium and an internal memory.
- the non-volatile storage medium stores an operating system and computer-readable instructions.
- the internal memory provides an environment for the operation of the operating system and computer-readable instructions in a non-volatile storage medium.
- the network interface of the computer device is used to communicate with an external terminal through a network connection.
- the computer-readable instructions are executed by a processor to implement a histogram-based iris image enhancement method.
- a computer device including a memory, a processor, and computer-readable instructions stored on the memory and executable on the processor.
- the processor implements the computer-readable instructions to implement the foregoing embodiments based on histograms.
- the steps of the iris image enhancement method in the figure are, for example, steps S10 to S50 shown in FIG. 2.
- the processor executes the computer-readable instructions, the functions of the modules / units of the iris image enhancement device based on the histogram in the foregoing embodiment are implemented, for example, modules 10 to 50 shown in FIG. 7. To avoid repetition, we will not repeat them here.
- One or more non-volatile readable storage media storing computer-readable instructions, which when executed by one or more processors, cause the one or more processors to execute the histogram-based Steps of the iris image enhancement method, or when the computer-readable instructions are executed by one or more processors, the functions of the modules / units of the iris image device based on the multiprocessing in the above embodiment are implemented. To avoid repetition, no longer here To repeat.
- Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Ophthalmology & Optometry (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
- Studio Devices (AREA)
- Facsimile Image Signal Circuits (AREA)
- Image Analysis (AREA)
Abstract
本申请公开了一种基于直方图的虹膜图像增强方法、装置、设备及存储介质。该基于直方图的虹膜图像增强方法包括:获取虹膜图像集;计算虹膜图像的对比度,基于同一用户标识,按照对比度由大到小的顺序对虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;基于每一用户标识从初始虹膜序列中依序获取预设数量的虹膜图像,组成初始虹膜集;采用自适应直方图均衡化算法对初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;对第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。该基于直方图的虹膜图像增强方法不仅提高了初始虹膜图像的整体对比度,增强内部细节,而且抑制了增强过程中放大的噪声,具有较好的增强效果,提高了虹膜图像的识别准确率。
Description
本申请以2018年5月25日提交的申请号为201810511993.7,名称为“基于直方图的虹膜图像增强方法、装置、设备及存储介质”的中国发明专利申请为基础,并要求其优先权。
本申请涉及图像处理领域,尤其涉及一种基于直方图的虹膜图像增强方法、装置、设备及存储介质。
虹膜作为一种重要的身份鉴别特征,具有唯一性、稳定性、可采集性和非侵犯性等特点。在虹膜识别系统中,经常需要清晰度较高的虹膜图像作为训练集,但是由于采集设备的限制和采集环境变化等因素的影响,都会导致采集的虹膜图像质量不佳,如对比度低和噪声干扰等问题都会影响虹膜纹理特征的凸显,进而影响虹膜图像训练集的清晰度和识别效率。为了提高识别的准确率,往往需要对虹膜图像进行增强处理,以凸显图像的纹理特征。目前通常只是对采集到的虹膜图像的对比度进行整体动态调整,然而经过如此处理的虹膜图像在识别系统中的准确率还是不高。
发明内容
基于此,有必要针对上述技术问题,提供一种基于直方图的虹膜图像增强方法、装置、设备及存储介质,以解决虹膜图像识别准确率不高的问题。
一种基于直方图的虹膜图像增强方法,包括:
获取虹膜图像集,所述虹膜图像集包括虹膜图像,所述虹膜图像包括用户标识;
计算所述虹膜图像集中的虹膜图像的对比度,按照对比度由大到小的顺序对虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;
从每个用户标识对应的初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集;
采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;
对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
一种基于直方图的虹膜图像增强装置,包括:
虹膜图像集获取模块,用于获取虹膜图像集,所述虹膜图像集包括虹膜图像,所述虹膜图像包括用户标识;
虹膜序列获取模块,用于计算所述虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;
初始虹膜集获取模块,用于从每个用户标识对应的初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集;
第一增强虹膜图像集获取模块,用于采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;
第二增强虹膜图像集获取模块,用于对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
一种计算机设备,包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机可读指令,所述处理器执行所述计算机可读指令时实现如下步骤:
获取虹膜图像集,所述虹膜图像集包括虹膜图像,所述虹膜图像包括用户标识;
计算所述虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对所述虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;
从每个用户标识对应的所述初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集;
采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;
对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
一个或多个存储有计算机可读指令的非易失性可读存储介质,所述计算机可读指令被一个或多个处理器执行时,使得所述一个或多个处理器执行如下步骤:
获取虹膜图像集,所述虹膜图像集包括虹膜图像,所述虹膜图像包括用户标识;
计算所述虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对所述虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;
从每个用户标识对应的所述初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集;
采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;
对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
本申请的一个或多个实施例的细节在下面的附图和描述中提出。本申请的其它特征和优点将从说明书、附图以及权利要求书变得更加明显。
为了更清楚地说明本申请实施例的技术方案,下面将对本申请实施例的描述中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动性的前提下,还可以根据这些附图获得其他的附图。
图1是本申请一实施例中基于直方图的虹膜图像增强方法的应用场景图;
图2是本申请实施例中基于直方图的虹膜图像增强方法的一流程图;
图3是图2中步骤S10的一具体实施方式的一流程图;
图4是图2中步骤S20的一具体实施方式的一流程图;
图5是图2中步骤S50的一具体实施方式的一流程图;
图6(a)是本申请实施例中一初始虹膜图像的示例图;
图6(b)是本申请实施例中一第二增强虹膜图像的示例图;
图7是本申请实施例中基于直方图的虹膜图像增强装置的一示意图;
图8是本申请实施例中计算机设备的一示意图。
下面将结合本申请实施例中的附图,对本申请实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例是本申请一部分实施例,而不是全部的实施例。基于本申请中的实施例,本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
本申请提供的基于直方图的虹膜图像增强方法,可以应用在计算机设备或系统中,用于对虹膜图像进行增强处理,以解决虹膜图像识别准确率不高的问题。其中,计算机设备可以但不限于各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备。
可选地,若该基于直方图的虹膜图像增强方法应用在系统中,该系统可以包括服务端和客户端。图1示出该基于直方图的虹膜图像增强方法应用在系统中的应用场景图。其中,服务端和客户端之间通过网络进行连接,客户端采集或者获取虹膜图像,服务端从客户端获取虹膜图像,客户端具体可以是摄像机、照相机、扫描仪或其他带有拍照功能的设备(手机或平板电脑等),或者存储有虹膜图像的存储设备。客户端可以为一个,也可以为复数个。服务端具体可以用一个服务器实现或者用复数个服务器组成的服务器集群实现。
在一实施例中,如图2所示,提供一种基于直方图的虹膜图像增强方法,以该方法应用在计算机设备中为例进行说明,包括以下步骤:
S10:获取虹膜图像集,虹膜图像集包括虹膜图像,虹膜图像包括用户标识。
其中,虹膜图像集是指由虹膜图像组成的图像集合,而虹膜图像是指通过摄像设备拍摄用户眼睛内部的虹膜所得到的图像。可选地,虹膜图像集可以是实时采集的,也可以预先存储在计算机设备中。一个虹膜图像集可以包括一个用户的虹膜图像,也可以包括多个用户的虹膜图像。例如,一个虹膜图像集中包括N个用户的虹膜图像,若每个用户都有M 幅虹膜图像,那么该虹膜图像集就有M*N幅虹膜图像。优选地,虹膜图像集中包括至少两幅虹膜图像。用户标识是指虹膜图像所属用户的标识,用于对虹膜图像按照所属用户进行分类,每一虹膜图像对应一用户标识,同一用户的虹膜图像对应相同的用户标识。
在一个具体实施方式中,可通过拍摄用户的眼睛获取预定数量的虹膜图像组成虹膜图像集,或者从计算机设备中获取预定数量的预先存储的虹膜图像组成虹膜图像集,或者通过拍摄用户的眼睛获取部分虹膜图像,再从计算机设备中获取另一部分预先存储的虹膜图像,两者共同组成虹膜图像集。
优选地,可选择在相同时间段内,采集若干用户的多幅虹膜图像作为虹膜图像集。例如,可以在阴雨天的中午采集,也可以在晴天的下午采集,这样能够避免获取的虹膜图像集中不同的虹膜图像因光线变化导致对比度相差较大的问题。
S20:计算虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列。
其中,对比度是一种衡量图像质量的指标,具体来说,虹膜图像的对比度是图像黑与白的比值,用于表征虹膜图像从黑到白的渐变层次。该比值越大,说明虹膜图像从黑到白的渐变层次越多,从而色彩表现越丰富。对比度对视觉效果的影响非常关键,一般来说对比度越大,图像越清晰醒目,色彩也越鲜明艳丽。对比度高的虹膜图像在一些暗部场景中的细节表现、清晰度和高速运动物体表现上优势更加明显。
初始虹膜序列,是指每个用户标识对应的虹膜图像按照对比度由大到小的顺序排列组成的虹膜序列。同一用户标识对应的虹膜图像按照对比度由大到小的顺序排列组成的虹膜序列为该用户标识对应的子虹膜序列。具体地,对每个用户标识的多幅虹膜图像一一进行对比度计算,并依据对比度由大到小的顺序对每个用户标识对应的多幅虹膜图像进行排序,得到每个用户标识对应的子虹膜序列。对比度越大的虹膜图像,清晰度也会越高,因此将虹膜图像集中每个用户标识对应的虹膜图像按照对比度从大到小的顺序排列。例如:初始虹膜序列中有N个用户标识,每个用户标识都包括M幅虹膜图像,则有N个子虹膜序列,且这N个子虹膜序列一起组成了初始虹膜序列,每个子虹膜序列中的虹膜图像是按照对比度由大到小的顺序排列的。
容易理解地,由于对比度是衡量图像质量的一种重要指标,因此,可通过计算比较每幅虹膜图像的对比度,并基于同一用户标识,按照对比度由大到小的顺序对虹膜图像集中的虹膜图像进行排序,进而将对比度较大的虹膜图像作为虹膜图像选取的标准,便于后续从中挑选对比度较大的虹膜图像进行处理,得到更具纹理特征的虹膜图像。
S30:从每个用户标识对应的初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集。
其中,预设数量是预先设定好的一个数值,用于选取一定数量的虹膜图像后续再进行增强处理。可选地,该预设数量可以根据训练样本量的要求来设定。例如:若在后续的模型训练中,要求每个用户标识的训练样本数量为P,则可以设定预设数量为P。可选地,可以根据每一用户标识对应的虹膜图像的数量来设置预设数量的值。例如,在每个用户标 识采集的图像为30幅的情况下,对应的预设数量可以设置为10幅,即只需要在这30幅虹膜图像中,按照对比度的大小顺序,在每个用户标识对应的初始虹膜序列中按照从大至小选取10幅虹膜图像作为初始虹膜集即可。初始虹膜集是从初始虹膜序列中选取的对比度数值靠前的预设数量的虹膜图像组成的集合。通过选取对比度较大的虹膜图像组成初始虹膜集,从而排除对比度较小的虹膜图像,减少了一些冗余图像,减轻后续虹膜图像增强处理的工作量,加快虹膜图像增强处理的速度,提高后续虹膜图像的处理效率。同时,由于选取对比度较大的虹膜图像作为初始虹膜集,能够提高虹膜图像的增强程度,进而提升虹膜图像的识别率。
S40:采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集。
其中,自适应直方图均衡化(Adaptive Histogram Equalization,AHE)算法采用移动子窗口技术,对包含初始虹膜图像像素的子窗口区域进行直方图均衡化,即求出该区域内的最低灰度值和最高灰度值,在该子窗口对应的区域内均衡化初始虹膜图像直方图,据此并按整幅初始虹膜图像直方图均衡化的灰度映射原理,给子窗口对应的区域内待处理图像像素重新赋值。容易理解地,初始虹膜图像灰度值整体偏低,其直方图分布在灰度值较小的区域,因此在低灰度区域相对较为集中,经过直方图均衡化后,增大了其灰度分布范围,且各个灰度级之间的距离也被增大,在整个灰度级上更为均匀,有效地扩展了初始虹膜图像的灰度分布范围,使得初始虹膜图像亮度增加,一定程度上提高了初始虹膜图像的对比度。
局部增强是指对图像的局部信息进行增强的处理过程,在一具体实施方式中,通过将整幅初始虹膜图像分成多个子块,利用直方图均衡化算法分散到初始虹膜图像的所有子块区域,通过子块区域直方图均衡化计算的叠加,自适应地增强虹膜图像的局部细节信息。因此,采用自适应直方图均衡化算法对初始虹膜图像进行局部增强后,使得初始虹膜图像尽可能地保留图像细节信息,局部纹理细节也能得到自适应调整,增强了图像的内部细节,整幅虹膜图像的细节信息更为清晰,从而可以得到纹理丰富清晰的第一增强虹膜图像集,可以提高后续的识别精度。
本实施例中,初始虹膜集中的初始虹膜图像的灰度值即为自适应直方图均衡化算法的输入,第一增强虹膜图像的灰度值为自适应直方图均衡化算法的输出。通过采用移动子窗口技术,对初始虹膜图像进行直方图均衡化,调节了虹膜图像灰度值的动态范围,通过对虹膜图像的局部进行自适应直方图均衡化,更好的凸显了局部特征,同时有效扩展初始虹膜图像像素灰度值的动态范围并细化了内部细节,使得虹膜图像的纹理细节更为明了。
S50:对第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
其中,第一增强虹膜图像是指对初始虹膜集中的初始虹膜图像采用自适应直方图均衡化算法进行增强处理后得到的虹膜图像。拉普拉斯算子(Laplacian operator)是一种二阶微分算子,适用于改善因为光线的漫反射造成的图像模糊。其原理是,在摄像记录图像 的过程中,光点将光漫反射到其周围区域,这种由于光的漫反射造成了图像一定程度的模糊,其模糊程度相对与正常情形下拍摄的图像来说,往往是拉普拉斯算子的常数倍,因此,对图像进行拉普拉斯算子锐化变换可以减少图像的模糊,提高图像的清晰度。因此,通过对第一增强虹膜图像进行锐化处理,突出第一增强虹膜图像的边缘细节特征,提高了第一增强虹膜图像的轮廓清晰度。
锐化处理是指对图像进行锐化的变换,用于加强图像中的目标边界和图像细节。第二增强虹膜图像是指对第一增强虹膜集中的虹膜图像采用拉普拉斯算子进行锐化处理后得到的虹膜图像。第一增强虹膜图像经过拉普拉斯算子锐化处理后,图像边缘细节特征被加强的同时也提高了第一增强虹膜图像的亮区边缘的清晰度,从而保护了第一增强虹膜图像的细节。
可选地,对第一增强虹膜图像采用拉普拉斯算子进行锐化的过程可以是:用拉普拉斯算子对第一增强虹膜图像像素的灰度值求二阶导数,二阶导数等于零处对应的像素就是图像的边缘像素,这样的处理可以将虹膜纹理边缘更加清晰的呈现出来,以便获取纹理细节更丰富的清晰的虹膜训练集,提高识别效果。
进一步地,在采用自适应直方图均衡化算法对初始虹膜进行增强处理的过程中,由于是通过移动子窗口的方式对初始虹膜图像进行直方图均衡化,为此会引入放大的噪声,产生干扰。因此,利用锐化处理中的拉普拉斯算子进行滤波,可以抑制由自适应直方图均衡化算法进行局部增强过程中引入的噪声,使得第二增强虹膜细节信息更为丰富。本实施例中,首先获取虹膜图像集,对虹膜图像集中的虹膜图像进行对比度计算,并提取出虹膜图像集中对比度较大的虹膜图像组成初始虹膜集,从而减少了质量不佳的虹膜图像,减少了冗余操作,有利于提高虹膜图像的增强程度和后续增强处理的效率。然后对初始虹膜集中的初始虹膜图像采用自适应直方图均衡化算法进行局部增强处理,调节了初始虹膜图像灰度值的动态范围,更好的凸显了局部特征,同时细化了内部细节,最后对增强后的第一增强虹膜图像进行锐化处理,以抑制由自适应直方图均衡化算法进行局部增强过程中引入的噪声,同时锐化处理后得到第二增强虹膜图像,保留了更多的虹膜图像的细节,整体上提高了虹膜图像的对比度,使虹膜图像的纹理特征更加清晰,以便得到纹理细节更丰富的清晰的虹膜训练集,提高后续识别的准确率。
在一实施例中,如图3所示,步骤S10中,即获取虹膜图像集,具体包括如下步骤:
S11:实时获取人眼和摄像头的实测距离,若实测距离不在距离阈值范围内,则发送提示消息。
其中,实测距离是指用户的眼睛距离摄像头的距离,距离阈值是指实验测定的过程中,通过反复测试给定的一个预设距离值,若被拍摄物在距离阈值处,计算机设备所采集的图像相比其他的位置处采集的图像质量更优。距离阈值范围是指在距离阈值上下所设定的界限,容易理解地,实测距离在距离阈值上下波动的一定范围内,也能拍摄到清晰的图像。在保证图像质量清晰的条件下,为了方便快速拍摄图像,可以设置该距离阈值±a%的数值范围作为距离阈值范围,可选地,a可以为5、10或15等。提示信息是用于提示用户进行 相应的调整以便拍摄清晰图像,提示信息包括但不限于箭头标识(如不同方向的箭头)、文字提示信息(如距离过远、距离适当或距离过近)和语音信息(如“请您靠近摄像头”、“正在采集”或“请您远离摄像头”)等。
获取到实测距离后,判断实测距离是否在距离阈值范围内,如果不在距离阈值范围内,发送提示信息,用户根据提示信息进行相应的调整,再获取实测距离,直到实测距离在距离阈值范围内为止,通过引导用户在距离阈值范围内,有利于提高后续获取的虹膜图像的质量。
在一个具体实施方式中,以红外摄像头为例,可以将红外摄像头的焦距作为距离阈值,具体地,可通过红外摄像头本身手动或自动对焦的方式进行调节,根据图像的清晰度确定该摄像头的焦距。
本实施例中,通过对实测距离和距离阈值范围的比较来发送对应的提示信息,可以引导用户快速地调整位置,提高虹膜图像采集的效率。
S12:若实测距离在距离阈值范围内,则控制摄像头进行连续拍摄,获取虹膜图像集。
可以理解地,当用户眼睛距离摄像头的距离在距离阈值范围内,这种情况下拍摄获取的虹膜质量较佳。连续拍摄可以获取同样一个人的多张虹膜图像,方便快捷,为后续的增强处理提供了质量较佳的虹膜图像集。
本实施例中,通过获取用户眼睛与摄像头的实测距离,并将实测距离与距离阈值范围进行比较,根据比较结果反馈对应的提示信息给用户,用户依据提示信息进行调整,当实测距离在阈值范围内,控制摄像头进行连续拍摄,获取虹膜图像,可以方便快捷地获取到虹膜图像,也提高了虹膜图像的质量。
在一实施例中,如图4所示,步骤S20中,即计算虹膜图像集中的虹膜图像的对比度,具体包括如下步骤:
S21:获取虹膜图像集中虹膜图像的每个像素的灰度值,并依次将每个像素作为中心像素。
其中,像素(Pixel)是数字图像的基本元素,像素是在模拟图像数字化时对连续空间进行离散化得到的。每个像素具有整数行(高)和整数列(宽)位置坐标,同时每个像素都具有整数灰度值或颜色值。一幅图像是由很多像素构成的。具体地,数字图像数据可以用矩阵来表示,因此可以采用矩阵理论和矩阵算法对数字图像进行分析和处理。灰度图像的像素信息就是一个矩阵,矩阵的行对应图像的高,矩阵的列对应图像的宽,矩阵元素对应图像的像素,矩阵元素的值就是像素的灰度值,它表示灰度图像中颜色的深度。具体地,可通过图像信息获取工具获取到虹膜图像的每个像素对应的灰度值。即给出图像对应的路径,通过路径读取到该路径下的图像。例如,可以通过imread函数实现:
I=imread('D:\lena.jpg');
其中,jpg为图像的格式,lean为图像的名称,“D:\”为lean图像的路径,I为lean图像对应的矩阵。中心像素是指在给定的区域内,位于中心位置的像素。本实施中,依次将每个像素作为中心像素是指在给定的区域内,将区域内的每个像素都作为中心像素。例 如,区域内有15个像素,这15个像素依次作为中心像素,那么就有15个中心像素。当边界的像素作为中心像素时,可通过扩展像素的方式将边界像素看作中心像素,即边界像素邻域内不存在的像素的灰度值设置成与该边界像素灰度值相等。例如一虹膜图像的矩阵为:
其中,第一行第一列的像素的灰度值为22,其左部和上部均不存在像素,那么在计算对比度时,将其左部和上部的像素的灰度值设置成与该边界像素相同的大小的灰度值,即左部和上部的灰度值均为22。
S22:根据预设邻域大小,计算每个中心像素的灰度值与对应邻域像素的灰度值之差。
其中,邻域像素是指与中心像素位置相邻的像素,例如,位于坐标(x,y)的像素p有两个水平和两个垂直的相邻像素,每个像素距(x,y)一个单位距离。坐标分别为:(x-1,y),(x+1,y),(x,y-1),(x,y+1)。此像素集合定义为像素p的4个邻域,用N4(p)表示。另外,像素p还有4个对角相邻像素,坐标为:(x-1,y-1),(x+1,y-1),(x-1,y+1),(x+1,y+1)。这四个对角相邻像素和N4(p)共同称为像素P的8邻域,用N8(P)表示。
若预设邻域大小取4,即取4邻域,则每个中心像素与对应邻域的像素的灰度值之差有4个,若中心像素的灰度值用h(x,y)表示,那么其与对应4邻域的像素的灰度值之差可通过如下公式得到:
q
1=h(x-1,y)-h(x,y);
q
2=h(x,y-1)-h(x,y);
q
3=h(x+1,y)-h(x,y);
q
4=h(x,y+1)-h(x,y);
容易理解地,当中心像素为边界像素时,其对应的灰度值之差q
1、q
2、q
3、q
4中至少有一个值为0。
S23:基于预设邻域大小和该虹膜图像对应矩阵的行数和列数,获取该虹膜图像中灰度值之差的个数。
若预设邻域大小为4,矩阵的行数和列数分别为m和n,则灰度值之差的个数k可通过如下公式得到:
k=4×(m-2)×(n-2)+3×(2×(m-2)+2×(n-2))+4×2;
若预设邻域大小为8,矩阵的行数和列数分别为m和n,则灰度值之差的个数k可通过如 下公式得到:
k=8×(m-2)×(n-2)+6×(2×(m-2)+2×(n-2))+4×3。
S24:将该虹膜图像中每个中心像素的灰度值与对应邻域像素的灰度值之差进行平方求和之后除以该虹膜图像中灰度值之差的个数,得到该虹膜图像的对比度。
其中,虹膜图像的对比度用C表示,灰度值之差分别为q
1、q
2…q
k,k为正整数。虹膜图像的对比度C的具体计算公式如下:
C=(q
1
2+q
2+…+q
k
2)/k;
由公式可知,对比度C是一个具体的数值。
本实施例中,首先获取虹膜图像的每个像素的灰度值,并依次将每个像素作为中心像素,根据预设邻域大小,计算中心像素的灰度值与预设邻域像素的灰度值差值,该灰度值差值的个数通过预设邻域的大小和虹膜图像对应矩阵的行数和列数计算得到后,将该虹膜图像中每个中心像素的灰度值与对应邻域像素的灰度值差值进行平方求和之后除以灰度值差值的个数,所得到结果即是该虹膜图像的对比度。通过步骤S21至步骤S24可以简单快速地计算出虹膜图像的对比度,还能通过比较对比度筛选出质量高的虹膜图像。
在一实施例中,步骤S40中,采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集,具体包括:
获取预设子图块,并依照预设子图块的位置顺序对初始虹膜图像中对应的像素块进行直方图均衡化处理;
其中,对初始虹膜图像中对应的像素块进行直方图均衡化处理,具体包括:
获取像素块的直方图中的每个灰度值的个数;
基于每个灰度值的个数计算每个灰度值的累积分布函数;
基于每个灰度值的累积分布函数对像素块进行灰度变换处理,得到均衡化后的像素块。
其中,预设子图块是指预先设定好的移动子窗口,用于作为对像素块进行对比度调整的标准。可以理解地,预设子图块越小,其包含的信息量越少,像素块的对比度拉伸程度越大。例如,预设子图块A的大小为32×32,预设子图块B的大小为8×8,则预设子图块B对应的像素块的对比度的拉伸程度大于预设子图块A对应的像素块的对比度的拉伸程度。像素块是指将初始虹膜图像进行划分获得的子图块。例如初始虹膜图像大小为64×64,将其按照2×2均分,则有4个像素块,每个像素块的大小为32×32。依照预设子图块的位置顺序是指按照子图块的每个像素块所在位置的顺序,如先从上到下再从左至右的顺序,或者先从左至右再从上到下的顺序。例如,一初始虹膜图像按照M×N均分,依照预设子图块的位置顺序可以是:第一行、第二行…第M行的顺序,也可以是第一列、第二列…第N列的顺序。
具体地,以预设子图块在初始虹膜图像的像素块上移动,并按像素块的位置顺序进行遍历,对像素块进行直方图均衡化,通过依照预设子图块的位置顺序对像素块进行操作,虹膜图像的局部信息得到自适应调整,增加了初始虹膜图像的内部细节。
图像的直方图是描述图像中各灰度值出现频率的统计关系,反映的是一幅图像中各个灰度值出现的次数。计算各个灰度值的累积分布函数是指获取到各个灰度值的个数后,将小于或者等于各个灰度值的个数累加后得到的分布函数,例如,N(n
1)、N(n
2)…N(n
k)为直方图中小于等于灰度值为n
k的个数,那么灰度值为r
k对应的累积分布函数为:N(r
k)=N(n
1)+N(n
2)+…N(n
k)。灰度变换处理具体方法是:将计算得到的累积分布函数作为变换函数,对初始虹膜图像中的各个像素块通过累积分布函数进行灰度变换,从而将灰度值相对较为集中的区域变换到了较大的区域,提高了初始虹膜图像中的像素块的对比度。
在一具体实施方式中,可通过图像信息工具获取像素块的直方图,然后统计像素块对应的直方图的各个灰度值的个数;计算各个灰度值的累积分布函数;基于累积分布函数对像素块进行灰度变换处理,得到均衡化后的像素块。通过累积变换函数进行灰度变换,将初始虹膜图像的像素块的直方图分布改变成均匀分布直方图分布,通过使用累积函数对灰度值进行调整实现了初始虹膜图像的对比度增强的效果。可以理解地,通过这三个步骤实现初始虹膜图像中的对应像素块的直方图均衡化。
本实施例中,将预设子图块在初始虹膜图像的像素块上依照子图块的位置顺序进行遍历后在进行直方图均衡化处理,使得初始虹膜图像的像素灰度值的动态范围得到有效扩展,并且局部信息也得到自适应调整,增强了内部细节。
在一实施例中,如图5所示,步骤S50中,即对第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,具体包括如下步骤:
S51:获取第一增强虹膜图像集中的第一增强虹膜图像的每个像素的灰度值,采用拉普拉斯算子对每个像素的灰度值都进行锐化,得到锐化后的像素灰度值。
具体地,可以直接读入第一增强虹膜图像集中的第一增强虹膜图像,获取每一虹膜图像像素灰度值,具体的读入方式和步骤S21类似,在此不再赘述。
基于二阶微分的拉普拉斯算子定义为:
对于第一增强虹膜图像R(x,y),其二阶导数为:
S52:基于第一增强虹膜图像中锐化后的像素灰度值,获取对应的第二增强虹膜图像。
将锐化后的像素灰度值替换原(x,y)像素处的灰度值得到第二增强虹膜图像。
如图6(a)和图6(b)所示,展示了对一幅初始虹膜图像(图6(a))进行自适应直方图均衡化算法增强和拉普拉斯算子锐化后的虹膜图像即第二增强虹膜图像(图6(b))的对比图。可以看出,初始虹膜图像的整体对比度较低,第二增强虹膜图像相对于初始虹膜图像,灰度值的动态范围得到有效调节,较好地呈现了6(a)中不清楚的局部信息,局部细节信息凸显(人眼图像中的虹膜图像视觉上变清晰),边缘细节比较丰富。
在本实施例中,首先获取第一增强虹膜图像集中的第一增强虹膜图像的每个像素的灰度值,对其进行拉普拉斯锐化处理,得到锐化后的像素灰度值后再获取对应的第二增强虹膜图像。将经过自适应直方图均衡化算法增强处理之后的虹膜图像采用拉普拉斯算子进行锐化,图像边缘细节特征被加强的同时抑制了第一增强虹膜图像增强过程中引入的噪声,从而保护了第一增强虹膜图像的细节。此外,上述步骤不仅简单方便,提高虹膜图像处理的实时性,而且处理后得到第二增强虹膜图像边缘细节特征更加突出明了,虹膜图像集的整体对比度得到较大提高,增强了虹膜图像的纹理特征,有利于提高虹膜图像的识别的准确率。
值得说明的是,为了验证该基于直方图的虹膜图像增强方法的有效性,按照本实施例中步骤S11和步骤S12的方法采集50个人眼虹膜图像,每个人眼6张共计600幅虹膜图像集,计算虹膜图像集的对比度,选取每个人眼对比度靠前的3幅,其中2幅用作训练,1幅用作验证。将此300张虹膜图像经过本实施例中的自适应直方图均衡化算法进行局部增强后,并将增强的虹膜图像采用本实施例中步骤S51至步骤S52的方法进行锐化处理,得到处理后的训练集和验证集。将未经处理训练集和经过处理训练集分别提取纹理特征,识别算法通过计算欧式距离或者通过支持向量机(Support Vector Machine,SVM)分类器来实现,计算比较识别率,作为该基于直方图的虹膜图像增强算法的增强效果。结果显示:未经处理的虹膜图像的识别率为83%,虹膜图像经本实施例中的基于直方图的虹膜图像增强方法处理后的识别率为99.3%,识别率提升了16.3%。
应理解,上述实施例中各步骤的序号的大小并不意味着执行顺序的先后,各过程的执行顺序应以其功能和内在逻辑确定,而不应对本申请实施例的实施过程构成任何限定。
在一实施例中,提供一种基于直方图的虹膜图像增强装置,该基于直方图的虹膜图像增强装置与上述实施例中基于直方图的虹膜图像增强方法一一对应。如图7所示,该基于直方图的虹膜图像增强装置包括虹膜图像集获取模块10、虹膜序列获取模块20、初始虹膜集获取模块30、第一增强虹膜图像集获取模块40和第二增强虹膜图像集获取模块50。其中,虹膜图像集获取模块10、虹膜序列获取模块20、初始虹膜集获取模块30、第一增 强虹膜图像集获取模块40和第二增强虹膜图像集获取模块50的实现功能与上述实施例中基于直方图的虹膜图像增强方法对应的步骤一一对应,为避免赘述,本实施例不一一详述。
虹膜图像集获取模块10,用于获取虹膜图像集,虹膜图像集包括虹膜图像,虹膜图像包括用户标识。
虹膜序列获取模块20,用于计算虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列。
初始虹膜集获取模块30,用于从每个用户标识对应的初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集。
第一增强虹膜图像集获取模块40,用于采用自适应直方图均衡化算法对初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集。
第二增强虹膜图像集获取模块50,用于对第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
具体地,虹膜图像集获取模块10包括实测距离检测单元11、虹膜图像集获取单元12。
实测距离检测单元11,用于实时获取人眼和摄像头的实测距离,若实测距离不在距离阈值范围内,则发送提示消息。
虹膜图像集获取单元12,用于若实测距离在距离阈值范围内,则控制摄像头进行连续拍摄,获取虹膜图像集。
具体地,虹膜序列获取模块20还包括对比度计算单元21,用于计算虹膜图像集中的虹膜图像的对比度。
具体地,对比度计算单元21包括灰度值获取子单元211、灰度值的差值获取子单元212、灰度值的差值个数获取子单元213和对比度计算子单元214。
灰度值获取子单元211,用于获取虹膜图像集中虹膜图像的每个像素的灰度值,并依次将每个像素作为中心像素。
灰度值的差值获取子单元212,用于根据预设邻域大小,计算每个中心像素的灰度值与对应邻域像素的灰度值之差。
灰度值的差值个数获取子单元213,用于基于预设邻域大小和该虹膜图像对应矩阵的行数和列数,获取该虹膜图像中灰度值之差的个数。
对比度计算子单元214,用于将该虹膜图像中每个中心像素的灰度值与对应邻域像素的灰度值之差进行平方求和之后除以该虹膜图像中灰度值之差的个数,得到该虹膜图像的对比度。
具体地,第一增强虹膜图像集获取模块40用于获取预设子图块,并依照预设子图块的位置顺序对初始虹膜图像中对应的像素块进行直方图均衡化处理;
其中,对初始虹膜图像中对应的像素块进行直方图均衡化处理,具体包括:
获取像素块的直方图中每个灰度值的个数;
基于每个灰度值的个数计算每个灰度值的累积分布函数;
基于每个灰度值的累积分布函数对像素块进行灰度变换处理,得到均衡化后的像素块。
具体地,第二增强虹膜图像集获取模块50包括锐化后灰度值获取单元51和第二增强虹膜图像获取单元52。
锐化后灰度值获取单元51,用于获取第一增强虹膜图像集中的第一增强虹膜图像的每个像素的灰度值,采用拉普拉斯算子对每个像素的灰度值都进行锐化,得到锐化后的像素灰度值。
第二增强虹膜图像获取单元52,用于基于第一增强虹膜图像中锐化后的像素灰度值,获取对应的第二增强虹膜图像。
关于基于直方图的虹膜图像增强装置的具体限定可以参见上文中对于基于直方图的虹膜图像增强方法的限定,在此不再赘述。上述基于直方图的虹膜图像增强装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中,也可以以软件形式存储于计算机设备中的存储器中,以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中,提供了一种计算机设备,该计算机设备可以是客户端,其内部结构图可以如图8所示。该计算机设备包括通过系统总线连接的处理器、存储器和网络接口。其中,该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质和内存储器。该非易失性存储介质存储有操作系统和计算机可读指令。该内存储器为非易失性存储介质中的操作系统和计算机可读指令的运行提供环境。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机可读指令被处理器执行时以实现一种基于直方图的虹膜图像增强方法。
在一个实施例中,提供了一种计算机设备,包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机可读指令,处理器执行计算机可读指令时实现上述实施例基于直方图的虹膜图像增强方法的步骤,例如图2所示的步骤S10至步骤S50。或者,处理器执行计算机可读指令时实现上述实施例基于直方图的虹膜图像增强装置的各模块/单元的功能,例如图7所示的模块10至模块50。为避免重复,这里不再赘述。
一个或多个存储有计算机可读指令的非易失性可读存储介质,该计算机可读指令被一个或多个处理器执行时,使得一个或多个处理器执行上述实施例中基于直方图的虹膜图像增强方法的步骤,或者,计算机可读指令被一个或多个处理器执行时实现上述实施例中基于多重处理的虹膜图像装置的各模块/单元的功能,为避免重复,这里不再赘述。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机可读指令来指令相关的硬件来完成,所述的计算机可读指令可存储于一非易失性计算机可读取存储介质中,该计算机可读指令在执行时,可包括如上述各方法的实施例的流程。其中,本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用,均可包括非易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。
所属领域的技术人员可以清楚地了解到,为了描述的方便和简洁,仅以上述各功能单 元、模块的划分进行举例说明,实际应用中,可以根据需要而将上述功能分配由不同的功能单元、模块完成,即将所述装置的内部结构划分成不同的功能单元或模块,以完成以上描述的全部或者部分功能。
以上所述实施例仅用以说明本申请的技术方案,而非对其限制;尽管参照前述实施例对本申请进行了详细的说明,本领域的普通技术人员应当理解:其依然可以对前述各实施例所记载的技术方案进行修改,或者对其中部分技术特征进行等同替换;而这些修改或者替换,并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围,均应包含在本申请的保护范围之内。
Claims (20)
- 一种基于直方图的虹膜图像增强方法,其特征在于,包括:获取虹膜图像集,所述虹膜图像集包括虹膜图像,所述虹膜图像包括用户标识;计算所述虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对所述虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;从每个用户标识对应的所述初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集;采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
- 如权利要求1所述的基于直方图的虹膜图像增强方法,其特征在于,所述获取虹膜图像集,包括:实时获取人眼和摄像头的实测距离,若所述实测距离不在距离阈值范围内,则发送提示消息;若所述实测距离在距离阈值范围内,则控制所述摄像头进行连续拍摄,获取所述虹膜图像集。
- 如权利要求1所述的基于直方图的虹膜图像增强方法,其特征在于,所述计算所述虹膜图像集中的虹膜图像的对比度,包括:获取所述虹膜图像集中虹膜图像的每个像素的灰度值,并依次将每个像素作为中心像素;根据预设邻域大小,计算每个中心像素的灰度值与对应邻域像素的灰度值之差;基于所述预设邻域大小和该虹膜图像对应矩阵的行数和列数,获取该虹膜图像中所述灰度值之差的个数;将该虹膜图像中每个中心像素的灰度值与对应邻域像素的灰度值之差进行平方求和之后除以该虹膜图像中所述灰度值之差的个数,得到该虹膜图像的对比度。
- 如权利要求1所述的基于直方图的虹膜图像增强方法,其特征在于,所述采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,包括:获取预设子图块,并依照所述预设子图块的位置顺序对初始虹膜图像中对应的像素块进行直方图均衡化处理;其中,对初始虹膜图像中对应的像素块进行直方图均衡化处理,包括:获取所述像素块的直方图中每个灰度值的个数;基于每个灰度值的个数计算每个灰度值的累积分布函数;基于每个灰度值的累积分布函数对像素块进行灰度变换处理,得到均衡化后的像素块。
- 如权利要求1所述的基于直方图的虹膜图像增强方法,其特征在于,所述对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,包括:获取所述第一增强虹膜图像集中的第一增强虹膜图像的每个像素的灰度值,采用拉普拉斯算子对每个像素的灰度值都进行锐化,得到锐化后的像素灰度值;基于第一增强虹膜图像中所述锐化后的像素灰度值,获取对应的第二增强虹膜图像。
- 一种基于直方图的虹膜图像增强装置,其特征在于,包括:虹膜图像集获取模块,用于获取虹膜图像集,所述虹膜图像集包括虹膜图像,所述虹膜图像包括用户标识;虹膜序列获取模块,用于计算所述虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;初始虹膜集获取模块,用于从每个用户标识对应的初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集;第一增强虹膜图像集获取模块,用于采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;第二增强虹膜图像集获取模块,用于对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
- 如权利要求6所述的基于直方图的虹膜图像增强装置,其特征在于,所述第一增强虹膜图像集获取模块,还用于获取预设子图块,并依照所述子图块的位置顺序对初始虹膜图像中对应的像素块进行直方图均衡化处理;其中,对初始虹膜图像中对应的像素块进行直方图均衡化处理,包括:获取所述像素块的直方图中每个灰度值的个数;基于每个灰度值的个数计算每个灰度值的累积分布函数;基于每个灰度值的累积分布函数对像素块进行灰度变换处理,得到均衡化后的像素块。
- 如权利要求6所述的基于直方图的虹膜图像增强装置,其特征在于,所述第二增强虹膜图像集获取模块,包括:锐化后灰度值获取单元,用于获取所述第一增强虹膜图像集中的第一增强虹膜图像的每个像素的灰度值,采用拉普拉斯算子对每个像素的灰度值都进行锐化,得到锐化后的像素灰度值;第二增强虹膜图像获取单元,用于基于第一增强虹膜图像中所述锐化后的像素灰度值,获取对应的第二增强虹膜图像。
- 如权利要求6所述的基于直方图的虹膜图像增强装置,其特征在于,所述虹膜图像集获取模块包括:实测距离检测单元,用于实时获取人眼和摄像头的实测距离,若所述实测距离不在距离阈值范围内,则发送提示消息;虹膜图像集获取单元,用于若所述实测距离在距离阈值范围内,则控制所述摄像头进 行连续拍摄,获取所述虹膜图像集。
- 如权利要求6所述的基于直方图的虹膜图像增强装置,其特征在于,所述虹膜序列获取模块包括:灰度值获取子单元,用于获取所述虹膜图像集中虹膜图像的每个像素的灰度值,并依次将每个像素作为中心像素;灰度值的差值获取子单元,用于根据预设邻域大小,计算每个中心像素的灰度值与对应邻域像素的灰度值之差;灰度值的差值个数获取子单元,用于基于所述预设邻域大小和该虹膜图像对应矩阵的行数和列数,获取该虹膜图像中所述灰度值之差的个数;对比度计算子单元,用于将该虹膜图像中每个中心像素的灰度值与对应邻域像素的灰度值之差进行平方求和之后除以该虹膜图像中所述灰度值之差的个数,得到该虹膜图像的对比度。
- 一种计算机设备,包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机可读指令,其特征在于,所述处理器执行所述计算机可读指令时实现如下步骤:获取虹膜图像集,所述虹膜图像集包括虹膜图像,所述虹膜图像包括用户标识;计算所述虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对所述虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;从每个用户标识对应的所述初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集;采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
- 如权利要求11所述的计算机设备,其特征在于,所述获取虹膜图像集,包括:实时获取人眼和摄像头的实测距离,若所述实测距离不在距离阈值范围内,则发送提示消息;若所述实测距离在距离阈值范围内,则控制所述摄像头进行连续拍摄,获取所述虹膜图像集。
- 如权利要求11所述的计算机设备,其特征在于,所述计算所述虹膜图像集中的虹膜图像的对比度,包括:获取所述虹膜图像集中虹膜图像的每个像素的灰度值,并依次将每个像素作为中心像素;根据预设邻域大小,计算每个中心像素的灰度值与对应邻域像素的灰度值之差;基于所述预设邻域大小和该虹膜图像对应矩阵的行数和列数,获取该虹膜图像中所述 灰度值之差的个数;将该虹膜图像中每个中心像素的灰度值与对应邻域像素的灰度值之差进行平方求和之后除以该虹膜图像中所述灰度值之差的个数,得到该虹膜图像的对比度。
- 如权利要求11所述的计算机设备,其特征在于,所述采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,包括:获取预设子图块,并依照所述预设子图块的位置顺序对初始虹膜图像中对应的像素块进行直方图均衡化处理;其中,对初始虹膜图像中对应的像素块进行直方图均衡化处理,包括:获取所述像素块的直方图中每个灰度值的个数;基于每个灰度值的个数计算每个灰度值的累积分布函数;基于每个灰度值的累积分布函数对像素块进行灰度变换处理,得到均衡化后的像素块。
- 如权利要求11所述的计算机设备,其特征在于,所述对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,包括:获取所述第一增强虹膜图像集中的第一增强虹膜图像的每个像素的灰度值,采用拉普拉斯算子对每个像素的灰度值都进行锐化,得到锐化后的像素灰度值;基于第一增强虹膜图像中所述锐化后的像素灰度值,获取对应的第二增强虹膜图像。
- 一个或多个存储有计算机可读指令的非易失性可读存储介质,其特征在于,所述计算机可读指令被一个或多个处理器执行时,使得所述一个或多个处理器执行如下步骤:获取虹膜图像集,所述虹膜图像集包括虹膜图像,所述虹膜图像包括用户标识;计算所述虹膜图像集中的虹膜图像的对比度,并按照对比度由大到小的顺序对所述虹膜图像集中每个用户标识对应的虹膜图像进行排序,得到每个用户标识对应的初始虹膜序列;从每个用户标识对应的所述初始虹膜序列中依据对比度由大到小的顺序获取预设数量的虹膜图像,组成初始虹膜集;采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,得到第一增强虹膜图像集;对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,得到第二增强虹膜图像集。
- 如权利要求16所述的非易失性可读存储介质,其特征在于,所述获取虹膜图像集,包括:实时获取人眼和摄像头的实测距离,若所述实测距离不在距离阈值范围内,则发送提示消息;若所述实测距离在距离阈值范围内,则控制所述摄像头进行连续拍摄,获取所述虹膜图像集。
- 如权利要求16所述的非易失性可读存储介质,其特征在于,所述计算所述虹膜图像集中的虹膜图像的对比度,包括:获取所述虹膜图像集中虹膜图像的每个像素的灰度值,并依次将每个像素作为中心像素;根据预设邻域大小,计算每个中心像素的灰度值与对应邻域像素的灰度值之差;基于所述预设邻域大小和该虹膜图像对应矩阵的行数和列数,获取该虹膜图像中所述灰度值之差的个数;将该虹膜图像中每个中心像素的灰度值与对应邻域像素的灰度值之差进行平方求和之后除以该虹膜图像中所述灰度值之差的个数,得到该虹膜图像的对比度。
- 如权利要求16所述的非易失性可读存储介质,其特征在于,所述采用自适应直方图均衡化算法对所述初始虹膜集中的初始虹膜图像进行局部增强处理,包括:获取预设子图块,并依照所述预设子图块的位置顺序对初始虹膜图像中对应的像素块进行直方图均衡化处理;其中,对初始虹膜图像中对应的像素块进行直方图均衡化处理,包括:获取所述像素块的直方图中每个灰度值的个数;基于每个灰度值的个数计算每个灰度值的累积分布函数;基于每个灰度值的累积分布函数对像素块进行灰度变换处理,得到均衡化后的像素块。
- 如权利要求16所述的非易失性可读存储介质,其特征在于,所述对所述第一增强虹膜图像集中的第一增强虹膜图像采用拉普拉斯算子进行锐化处理,包括:获取所述第一增强虹膜图像集中的第一增强虹膜图像的每个像素的灰度值,采用拉普拉斯算子对每个像素的灰度值都进行锐化,得到锐化后的像素灰度值;基于第一增强虹膜图像中所述锐化后的像素灰度值,获取对应的第二增强虹膜图像。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810511993.7 | 2018-05-25 | ||
CN201810511993.7A CN108846319A (zh) | 2018-05-25 | 2018-05-25 | 基于直方图的虹膜图像增强方法、装置、设备及存储介质 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2019223069A1 true WO2019223069A1 (zh) | 2019-11-28 |
Family
ID=64213519
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2018/094397 WO2019223069A1 (zh) | 2018-05-25 | 2018-07-04 | 基于直方图的虹膜图像增强方法、装置、设备及存储介质 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN108846319A (zh) |
WO (1) | WO2019223069A1 (zh) |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111861947A (zh) * | 2020-06-24 | 2020-10-30 | 江苏理工学院 | 提高直方图技术增强图像的信息熵的方法、装置 |
CN112561806A (zh) * | 2020-11-13 | 2021-03-26 | 连云港杰瑞自动化有限公司 | 一种微电阻率扫描测井仪图像增强方法及系统 |
CN113487525A (zh) * | 2021-07-06 | 2021-10-08 | 河南慧联世安信息技术有限公司 | 一种基于双平台直方图的自迭代红外图像增强方法 |
CN114298916A (zh) * | 2021-11-11 | 2022-04-08 | 电子科技大学 | 一种基于灰度拉伸和局部增强的X-Ray图像增强方法 |
CN114820397A (zh) * | 2022-07-01 | 2022-07-29 | 托伦斯半导体设备启东有限公司 | 基于关注程度的图像增强方法及系统 |
CN114926374A (zh) * | 2022-07-21 | 2022-08-19 | 四川新迎顺信息技术股份有限公司 | 一种基于ai的图像处理方法、装置、设备及可读存储介质 |
CN115170446A (zh) * | 2022-09-06 | 2022-10-11 | 聊城市洛溪信息科技有限公司 | 基于形态学处理的自适应钣金图像增强方法 |
CN115456917A (zh) * | 2022-11-11 | 2022-12-09 | 中国石油大学(华东) | 有益于目标准确检测的图像增强方法、装置、设备及介质 |
CN115761604A (zh) * | 2023-01-10 | 2023-03-07 | 矿冶科技集团有限公司 | 炉口开闭状态识别方法及装置 |
CN111242879B (zh) * | 2020-01-17 | 2023-05-16 | 郑州大学 | 图像处理方法、装置、电子设备以及介质 |
CN116137022A (zh) * | 2023-04-20 | 2023-05-19 | 山东省三河口矿业有限责任公司 | 一种用于地下采矿远程监控的数据增强方法 |
CN116309203A (zh) * | 2023-05-19 | 2023-06-23 | 中国人民解放军国防科技大学 | 一种偏振视觉自适应增强的无人平台运动估计方法和装置 |
CN116485686A (zh) * | 2023-06-19 | 2023-07-25 | 青岛国源中创电气自动化工程有限公司 | 一种活性污泥法的污水处理图像增强方法 |
CN116342588B (zh) * | 2023-05-22 | 2023-08-11 | 徕兄健康科技(威海)有限责任公司 | 一种脑血管图像增强方法 |
CN116758059A (zh) * | 2023-08-10 | 2023-09-15 | 吉林交通职业技术学院 | 一种用于路基路面的视觉无损检测方法 |
CN116863323A (zh) * | 2023-09-04 | 2023-10-10 | 济宁鑫惠生水产养殖专业合作社 | 一种用于渔业养殖水源污染的视觉检测方法和系统 |
CN116934755A (zh) * | 2023-09-18 | 2023-10-24 | 中国人民解放军总医院第八医学中心 | 基于直方图均衡化的肺结核ct图像增强系统 |
CN117011302A (zh) * | 2023-10-08 | 2023-11-07 | 山东济宁运河煤矿有限责任公司 | 一种基于煤矸识别的智能干选系统 |
CN117593193A (zh) * | 2024-01-19 | 2024-02-23 | 山东海天七彩建材有限公司 | 一种基于机器学习的钣金图像增强方法及系统 |
CN117635718A (zh) * | 2024-01-24 | 2024-03-01 | 泰安万川电器设备有限公司 | 一种基于图像增强的弱光环境下矿车定位方法及系统 |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111246010B (zh) * | 2019-04-24 | 2020-11-10 | 吕衍荣 | 基于信号分析的环境监测方法 |
CN110222763B (zh) * | 2019-06-04 | 2021-08-03 | 浙江大华技术股份有限公司 | 一种直方图匹配方法、移动终端和计算机存储介质 |
CN111143589A (zh) * | 2019-12-06 | 2020-05-12 | Oppo广东移动通信有限公司 | 一种图像处理方法及装置、存储介质 |
CN111340767B (zh) * | 2020-02-21 | 2023-12-12 | 四川大学华西医院 | 一种脑部肿瘤头皮定位图像处理方法及系统 |
CN111915541B (zh) * | 2020-07-31 | 2021-08-17 | 平安科技(深圳)有限公司 | 基于人工智能的图像增强处理方法、装置、设备及介质 |
CN116563172B (zh) * | 2023-07-11 | 2023-09-08 | 天津智教云科技有限公司 | 一种vr全球化在线教育互动优化增强方法及装置 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20030066512A (ko) * | 2003-07-04 | 2003-08-09 | 김재민 | 노이즈에 강인한 저용량 홍채인식 시스템 |
CN101231747A (zh) * | 2007-01-22 | 2008-07-30 | 清华大学 | 一种x射线图像边缘和对比度综合增强方法 |
KR20080085975A (ko) * | 2007-03-21 | 2008-09-25 | 한국과학기술원 | 얼굴인식 전처리장치 및 방법과 이를 이용한얼굴인식시스템 |
CN103136530A (zh) * | 2013-02-04 | 2013-06-05 | 国核自仪系统工程有限公司 | 在复杂工业环境下的视频图像中自动识别目标图像的方法 |
CN106548113A (zh) * | 2015-09-16 | 2017-03-29 | 上海市公安局刑事侦查总队 | 图像识别方法及系统 |
CN108062746A (zh) * | 2016-11-09 | 2018-05-22 | 深圳市优朋普乐传媒发展有限公司 | 一种视频图像处理方法与装置、视频编码系统 |
-
2018
- 2018-05-25 CN CN201810511993.7A patent/CN108846319A/zh not_active Withdrawn
- 2018-07-04 WO PCT/CN2018/094397 patent/WO2019223069A1/zh active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20030066512A (ko) * | 2003-07-04 | 2003-08-09 | 김재민 | 노이즈에 강인한 저용량 홍채인식 시스템 |
CN101231747A (zh) * | 2007-01-22 | 2008-07-30 | 清华大学 | 一种x射线图像边缘和对比度综合增强方法 |
KR20080085975A (ko) * | 2007-03-21 | 2008-09-25 | 한국과학기술원 | 얼굴인식 전처리장치 및 방법과 이를 이용한얼굴인식시스템 |
CN103136530A (zh) * | 2013-02-04 | 2013-06-05 | 国核自仪系统工程有限公司 | 在复杂工业环境下的视频图像中自动识别目标图像的方法 |
CN106548113A (zh) * | 2015-09-16 | 2017-03-29 | 上海市公安局刑事侦查总队 | 图像识别方法及系统 |
CN108062746A (zh) * | 2016-11-09 | 2018-05-22 | 深圳市优朋普乐传媒发展有限公司 | 一种视频图像处理方法与装置、视频编码系统 |
Cited By (31)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111242879B (zh) * | 2020-01-17 | 2023-05-16 | 郑州大学 | 图像处理方法、装置、电子设备以及介质 |
CN111861947B (zh) * | 2020-06-24 | 2023-09-22 | 江苏理工学院 | 提高直方图技术增强图像的信息熵的方法、装置 |
CN111861947A (zh) * | 2020-06-24 | 2020-10-30 | 江苏理工学院 | 提高直方图技术增强图像的信息熵的方法、装置 |
CN112561806A (zh) * | 2020-11-13 | 2021-03-26 | 连云港杰瑞自动化有限公司 | 一种微电阻率扫描测井仪图像增强方法及系统 |
CN113487525A (zh) * | 2021-07-06 | 2021-10-08 | 河南慧联世安信息技术有限公司 | 一种基于双平台直方图的自迭代红外图像增强方法 |
CN114298916A (zh) * | 2021-11-11 | 2022-04-08 | 电子科技大学 | 一种基于灰度拉伸和局部增强的X-Ray图像增强方法 |
CN114298916B (zh) * | 2021-11-11 | 2023-04-18 | 电子科技大学 | 一种基于灰度拉伸和局部增强的X-Ray图像增强方法 |
CN114820397A (zh) * | 2022-07-01 | 2022-07-29 | 托伦斯半导体设备启东有限公司 | 基于关注程度的图像增强方法及系统 |
CN114926374A (zh) * | 2022-07-21 | 2022-08-19 | 四川新迎顺信息技术股份有限公司 | 一种基于ai的图像处理方法、装置、设备及可读存储介质 |
CN115170446A (zh) * | 2022-09-06 | 2022-10-11 | 聊城市洛溪信息科技有限公司 | 基于形态学处理的自适应钣金图像增强方法 |
CN115170446B (zh) * | 2022-09-06 | 2022-11-18 | 聊城市洛溪信息科技有限公司 | 基于形态学处理的自适应钣金图像增强方法 |
CN115456917A (zh) * | 2022-11-11 | 2022-12-09 | 中国石油大学(华东) | 有益于目标准确检测的图像增强方法、装置、设备及介质 |
CN115761604A (zh) * | 2023-01-10 | 2023-03-07 | 矿冶科技集团有限公司 | 炉口开闭状态识别方法及装置 |
CN116137022A (zh) * | 2023-04-20 | 2023-05-19 | 山东省三河口矿业有限责任公司 | 一种用于地下采矿远程监控的数据增强方法 |
CN116137022B (zh) * | 2023-04-20 | 2023-08-22 | 山东省三河口矿业有限责任公司 | 一种用于地下采矿远程监控的数据增强方法 |
CN116309203A (zh) * | 2023-05-19 | 2023-06-23 | 中国人民解放军国防科技大学 | 一种偏振视觉自适应增强的无人平台运动估计方法和装置 |
CN116342588B (zh) * | 2023-05-22 | 2023-08-11 | 徕兄健康科技(威海)有限责任公司 | 一种脑血管图像增强方法 |
CN116485686A (zh) * | 2023-06-19 | 2023-07-25 | 青岛国源中创电气自动化工程有限公司 | 一种活性污泥法的污水处理图像增强方法 |
CN116485686B (zh) * | 2023-06-19 | 2023-08-29 | 青岛国源中创电气自动化工程有限公司 | 一种活性污泥法的污水处理图像增强方法 |
CN116758059B (zh) * | 2023-08-10 | 2023-10-20 | 吉林交通职业技术学院 | 一种用于路基路面的视觉无损检测方法 |
CN116758059A (zh) * | 2023-08-10 | 2023-09-15 | 吉林交通职业技术学院 | 一种用于路基路面的视觉无损检测方法 |
CN116863323B (zh) * | 2023-09-04 | 2023-11-24 | 济宁鑫惠生水产养殖专业合作社 | 一种用于渔业养殖水源污染的视觉检测方法和系统 |
CN116863323A (zh) * | 2023-09-04 | 2023-10-10 | 济宁鑫惠生水产养殖专业合作社 | 一种用于渔业养殖水源污染的视觉检测方法和系统 |
CN116934755A (zh) * | 2023-09-18 | 2023-10-24 | 中国人民解放军总医院第八医学中心 | 基于直方图均衡化的肺结核ct图像增强系统 |
CN116934755B (zh) * | 2023-09-18 | 2023-12-01 | 中国人民解放军总医院第八医学中心 | 基于直方图均衡化的肺结核ct图像增强系统 |
CN117011302A (zh) * | 2023-10-08 | 2023-11-07 | 山东济宁运河煤矿有限责任公司 | 一种基于煤矸识别的智能干选系统 |
CN117011302B (zh) * | 2023-10-08 | 2024-01-09 | 山东济宁运河煤矿有限责任公司 | 一种基于煤矸识别的智能干选系统 |
CN117593193A (zh) * | 2024-01-19 | 2024-02-23 | 山东海天七彩建材有限公司 | 一种基于机器学习的钣金图像增强方法及系统 |
CN117593193B (zh) * | 2024-01-19 | 2024-04-23 | 山东海天七彩建材有限公司 | 一种基于机器学习的钣金图像增强方法及系统 |
CN117635718A (zh) * | 2024-01-24 | 2024-03-01 | 泰安万川电器设备有限公司 | 一种基于图像增强的弱光环境下矿车定位方法及系统 |
CN117635718B (zh) * | 2024-01-24 | 2024-04-26 | 泰安万川电器设备有限公司 | 一种基于图像增强的弱光环境下矿车定位方法及系统 |
Also Published As
Publication number | Publication date |
---|---|
CN108846319A (zh) | 2018-11-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2019223069A1 (zh) | 基于直方图的虹膜图像增强方法、装置、设备及存储介质 | |
WO2019223068A1 (zh) | 虹膜图像局部增强方法、装置、设备及存储介质 | |
WO2019153739A1 (zh) | 基于人脸识别的身份认证方法、装置、设备和存储介质 | |
WO2017076040A1 (zh) | 应用于连拍过程中的图像处理方法和装置 | |
US9639956B2 (en) | Image adjustment using texture mask | |
US9483835B2 (en) | Depth value restoration method and system | |
JP6255486B2 (ja) | 情報認識のための方法及びシステム | |
CN106056064A (zh) | 一种人脸识别方法及人脸识别装置 | |
US9881202B2 (en) | Providing visual effects for images | |
US20150055824A1 (en) | Method of detecting a main subject in an image | |
CN105243371A (zh) | 一种人脸美颜程度的检测方法、系统及拍摄终端 | |
CN107346546B (zh) | 一种图像处理方法及装置 | |
CN111383232A (zh) | 抠图方法、装置、终端设备及计算机可读存储介质 | |
WO2019223066A1 (zh) | 虹膜图像全局增强方法、装置、设备及存储介质 | |
CN112418150B (zh) | 掌静脉图像评估方法、装置、计算机设备及存储介质 | |
WO2019223067A1 (zh) | 基于多重处理的虹膜图像增强方法、装置、设备及介质 | |
US8983188B1 (en) | Edge-aware smoothing in images | |
CN111145086A (zh) | 一种图像处理方法、装置及电子设备 | |
WO2022135574A1 (zh) | 肤色检测方法、装置、移动终端和存储介质 | |
WO2021189770A1 (zh) | 基于人工智能的图像增强处理方法、装置、设备及介质 | |
RU2697627C1 (ru) | Способ корректировки освещенности объекта на изображении в последовательности изображений и вычислительное устройство пользователя, реализующее упомянутый способ | |
WO2017177616A1 (zh) | 一种人脸识别方法及装置、画面显示方法及装置 | |
CN117496019B (zh) | 一种驱动静态图像的图像动画处理方法及系统 | |
RU2608239C1 (ru) | Способ и система определения пригодности изображения документа для оптического распознавания символов и других операций по обработке изображений | |
CN107153806B (zh) | 一种人脸检测方法及装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18919727 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 11/03/2021) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 18919727 Country of ref document: EP Kind code of ref document: A1 |