WO2021102913A1 - 图像处理方法、装置及存储介质 - Google Patents
图像处理方法、装置及存储介质 Download PDFInfo
- Publication number
- WO2021102913A1 WO2021102913A1 PCT/CN2019/121987 CN2019121987W WO2021102913A1 WO 2021102913 A1 WO2021102913 A1 WO 2021102913A1 CN 2019121987 W CN2019121987 W CN 2019121987W WO 2021102913 A1 WO2021102913 A1 WO 2021102913A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- pixel
- gray value
- image
- target noise
- denoised
- Prior art date
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 29
- 238000000034 method Methods 0.000 claims abstract description 39
- 238000004590 computer program Methods 0.000 claims description 12
- 238000001914 filtration Methods 0.000 claims description 9
- 230000000694 effects Effects 0.000 abstract description 10
- 238000001514 detection method Methods 0.000 description 15
- 238000004519 manufacturing process Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 230000004044 response Effects 0.000 description 4
- 230000009466 transformation Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000005670 electromagnetic radiation Effects 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000001931 thermography Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
Definitions
- This application relates to the field of image processing technology, and in particular to an image processing method, device and storage medium.
- the images collected by the image sensor usually contain some fixed patterns of noise, and these noises appear in a fixed position on each image collected by the image sensor.
- the infrared sensor due to the limitation of the manufacturing process, the response characteristics of the detection units on the infrared focal plane array are inconsistent, and the detection units have non-uniformities, resulting in some fixed patterns of noise in the final collected image.
- the existence of noise will seriously affect the clarity and display effect of the image, so it is necessary to denoise the image.
- denoising images in related technologies fixed pattern noise and actual scene objects cannot be effectively distinguished, resulting in unsatisfactory denoising effects. For example, vertical stripes and edges of vertical objects cannot be effectively judged, which will result in vertical object edges. Artificial flaws in vertical form. Therefore, it is necessary to improve the image noise removal method to improve the image denoising effect.
- this application provides an image processing method, device and storage medium.
- an image processing method including:
- an image processing apparatus including a processor, a memory, and a computer program stored on the memory, and the processor implements the following steps when the processor executes the computer program:
- a computer-readable storage medium having a computer program stored on the computer-readable storage medium, and when the computer program is executed by a processor, the image processing described in any one of the present application is implemented method.
- the calibration data used to determine the target noise frequency band and gray value is determined in advance, and then the gray value of the target noise contained in each pixel in the image to be denoised is determined according to the calibration , According to the gray value of the target noise contained in each pixel, the image to be denoised is denoised.
- the noise and real objects can be effectively identified from the image to be denoised, and the gray value of the noise can be accurately estimated to achieve a better denoising effect and improve the accuracy of infrared denoising.
- Fig. 1 is an image including vertical fringe noise provided by an embodiment of the present invention.
- Fig. 2 is a flowchart of an image processing method provided by an embodiment of the present invention.
- Fig. 3 is a block diagram of the logical structure of an image denoising device provided by an embodiment of the present invention.
- Fig. 4 is a block diagram of the logical structure of another image denoising device provided by an embodiment of the present invention.
- the images collected by the image sensor usually contain some fixed patterns of noise, and these noises appear on each image collected by the image sensor.
- the infrared sensor due to the limitation of the manufacturing process, the response characteristics of the detection units on the infrared focal plane array are inconsistent, and the detection units have non-uniformities, resulting in some fixed patterns of noise in the final collected image. For example, if the detection units in the same row of the infrared focal plane array share an output circuit, due to the difference in the bias voltage of the output circuit of each row, the gray values of the pixels in the adjacent two rows will be obviously different, and the final image will be collected. Horizontal fringe noise will appear.
- the detection units in the same column of the infrared focal plane array share one output circuit, then due to the difference in the bias voltage of the output circuit of each column, a large amount of vertical fringe noise will appear on the image.
- the output circuits of the detection units of the infrared focal plane array are shared according to other modes or one detection unit independently uses one output circuit, then fixed noises of other modes will appear. As shown in Figure 1, the collected image contains many vertical stripes of noise.
- the related technology denoises the image
- some use the frequency domain high-pass filtering method that is, the image data is converted from the spatial domain to the frequency domain. Because the noise is often at high frequency, it can filter out the high frequency part, according to the high frequency part.
- the gray value of the pixel points to obtain the gray value of the noise, and then the image is denoised according to the gray value of the noise.
- this method sometimes cannot effectively distinguish between noise and objects in the real scene, especially when the real object itself is quite different, for example, the real object is an object with more edges and corners, then it is more difficult to distinguish the noise from the real object. Edges, such as indistinguishable vertical stripes and object edges. In this case, some defects will be produced after the method is used to denoise the image noise, and the denoising effect is not ideal.
- Some technologies will set a shielded area in the image sensor.
- the detection unit in this shielded area cannot be light-sensitive. Therefore, the output value of each detection unit in this area reflects the difference in the response of the detection unit, which can then be used as a reference to determine the factor. Detect the noise generated by the response difference of the unit, and perform denoising processing on the image. For example, each row or column retains one or more detection units that are blocked and cannot be exposed to light, so the noise of the row or column can be determined according to the output value of the blocked detection unit.
- this method requires the image image sensor to reserve a block area, which wastes the space of the sensor and also increases the difficulty of the manufacturing process of the sensor.
- this application provides an image processing method. Since the noise is a fixed pattern of noise, the calibration data can be determined by the noise on the reference image. The calibration data is used to determine the reference frequency band and reference gray value of the noise, and then According to the calibration data, the noise is effectively identified from the image to be denoised, and the denoising process is performed to achieve a better denoising effect. Specifically, as shown in FIG. 2, the method may include the following steps:
- S204 Determine the gray value of the target noise contained in each pixel of the image to be denoised according to the predetermined calibration data; wherein the calibration data is obtained based on the target noise in the reference image, and is used to determine the target The frequency band of the noise and the gray value of the target noise;
- S206 Perform denoising processing on the image to be denoised according to the gray value of the target noise contained in each pixel.
- the image processing method of this application can be used in various image acquisition devices, such as an infrared thermal imager.
- the image acquisition device can directly perform denoising processing after acquiring an image, and of course it can also be used in other image acquisition devices.
- the electronic equipment acquires and collects the image to be denoised from the image acquisition device, and then performs denoising processing.
- the gray value of a pixel may refer to a temperature value.
- the image to be denoised in this application may be various remote sensing images.
- the remote sensing image is an image obtained by receiving electromagnetic radiation information of the detected target.
- the image to be denoised may be collected by an infrared sensor.
- Infrared image of course, the image to be denoised can also be an image obtained through other electromagnetic waves, which is not limited in this application.
- the target noise in this application is a fixed pattern of noise. Since the main cause of the noise is the deviation of the output circuit of the detection unit, for the image collected by the same sensor, the position of the noise in the image is basically fixed.
- the fixed pattern noise includes various fringe noises.
- the target noise may be horizontal fringe noise, vertical fringe noise, or both horizontal and vertical fringe noise.
- the target noise may also be other noises, such as shadows, ghost points, etc., that appear at a fixed position in the image.
- the main cause of noise is the deviation of the output circuit of the detection unit, the gray value of some pixels in the image is suddenly higher or lower than that of neighboring pixels. Therefore, the noise is usually contained in the sharp change of the gray value. pixel. Therefore, the high frequency part can be filtered out by the frequency domain filtering method to obtain noise. However, this method tends to treat large changes in the real object as noise.
- the target noise contained in a reference image can be used to determine the calibration data.
- the calibration data is used to determine the frequency band of the target noise in the image to be denoised and the gray value of the target noise, so that The gray value of the determined target noise is more accurate.
- the reference image may be an image of a planar object collected by an image sensor, where the image sensor is a sensor that collects the image to be denoised. Since the surface of the plane object is basically the same, there are no sharp places such as edges and corners, so the pixels with larger differences in the gray value of the adjacent pixels on the collected image can basically be considered to be caused by noise, so the image is displayed The part where the gray value changes sharply is noise, not the real object itself. Therefore, after the reference image is collected, the reference image can be Fourier transformed or a pre-designed high-pass filter can be used to filter the reference image to obtain the calibration data.
- the calibration data may include the reference frequency band of the target noise, the reference gray value of the target noise, and the reference variance of the gray value of the target noise.
- the reference frequency band, reference gray value and reference variance are all obtained based on the target noise in the reference image, and are used as a reference when denoising the image subsequently.
- the calibration data may also only include the frequency band and gray value of the target noise.
- the reference gray value of the target noise may be the average value of the noise gray value of the entire image, or the average value of the noise gray value of each row or each column, which can be specifically set according to actual requirements.
- the Fourier transform of the reference image can be performed to obtain the spectrogram corresponding to the reference image.
- one-dimensional Fourier transform can be performed on the reference image, for example, for each row or each column of image data.
- Fourier transformation is used to obtain the spectrogram of each row or column of image data, and then the frequency band of the target noise and the average gray value of the noise of each row or column can be determined according to the spectrogram of each row or column, as the Refer to the gray value, and then calculate the variance of a gray value according to the average gray value of each row or each column, as the reference variance.
- a two-dimensional Fourier transform on the image data of the reference image to obtain a spectrogram of the entire image, and then determine the frequency band of the target noise and the average value of the noise gray value of the entire image according to the spectrogram.
- a pre-designed high-pass filter may be used to filter the reference image to obtain the reference gray value of the reference frequency band and the reference variance. The specific method used to determine the calibration data based on the reference image can be determined according to the actual situation, and this application is not limited.
- the image to be denoised collected by the image sensor can be obtained.
- the reference image is an original image without contrast stretching
- the image to be denoised may also be an original image collected by the sensor without contrast stretching.
- the gray value of each pixel in the image to be denoised containing the target noise can be determined according to the predetermined calibration data, and the denoising process is performed on the image to be denoised according to the gray value of each pixel containing the noise.
- the reference frequency band of the target noise in the calibration data can be determined from the image to be denoised which contains the target noise, and the pixel that does not contain the target noise. What are the pixel points, and what is the gray value of the target noise in the pixel points containing the target noise.
- the Fourier transform of the image to be denoised can be performed first to obtain the spectrogram of the image to be denoised, or it can be pre-designed high-pass filtering.
- the device performs filtering processing on the image to be denoised, and then determines whether each pixel contains the target noise according to the reference frequency band in the calibration data.
- the image data of each row or column in the denoising image can be Fourier changed to obtain the spectrogram of each row or column of image data, and then the spectrogram of each row or column and the calibration data can be Refer to the frequency band to determine whether the row or column contains the target noise.
- the first pixel after performing Fourier transformation on the image to be denoised, or filtering with a high-pass filter, the first pixel can be determined from the image to be denoised according to the reference frequency band in the calibration data.
- a pixel is a pixel outside the reference frequency band. For the pixel outside the reference frequency band, it can be considered that it does not contain the target noise. Therefore, the gray value of these pixels containing the target noise can be set to 0.
- the pixel with a frequency greater than 10KHZ can be considered to contain the target noise pixel, if the frequency is less than 10KHZ, it is considered that this pixel does not contain the target noise Therefore, the gray value of this pixel containing the target noise is 0.
- the second pixel can be determined from the image to be denoised according to the reference frequency band in the calibration data.
- Two pixels are the pixels within the reference frequency band. For the pixels within the reference frequency band, these pixels can be considered to contain the target noise. Therefore, these pixels can be further determined based on the reference gray value and the reference variance in the calibration data Contains the gray value of the target noise.
- the pixels with large gray value changes in the image to be denoised may be the result of the difference between the noise and the different parts of the real object.
- the distance between the gray value of the second pixel and the reference gray value can be determined, and then the distance and the reference variance are determined.
- the probability that each second pixel contains the target noise Since the reference gray value determined according to the reference image can be the average value of the gray value of each row or each column containing the target noise, the reference variance is also obtained from the average gray value of the variance of each row or each column, which explains the gray value of the target noise The value can fluctuate within the reference variance. Assuming that the reference gray value is 10 and the reference variance is 2, it means that the gray value of the target noise can be between 8-12.
- the distance between the gray value of the second pixel and the reference gray value can be determined, and then the distance is compared with the variance. If the distance is closer to the variance, the more likely the second pixel contains the target noise. Larger, the smaller the distance and variance, the smaller the probability that the second pixel contains the target noise.
- the distance may be one or more of difference, norm distance, Euclidean distance, Manhattan distance, Hamming distance, or cosine distance.
- the norm distance can be a two-norm distance or a distance in various other norm spaces.
- the reference gray value of the target noise is 10
- the gray value of the second pixel is 8. Therefore, one or more of the above-mentioned distances between the two gray values can be calculated, and then compared with the variance , In order to determine the probability that each second pixel contains the target noise.
- the gray value of the second pixel containing the target noise can be determined according to the gray value of the second pixel and the probability that the second pixel contains the target noise.
- the gray value of the second pixel containing the target noise can be determined according to the gray value of the second pixel and the designated pixel and the probability of containing the target noise. For example, the gray value of the second pixel can be multiplied by the corresponding probability, and the gray value of the designated pixel can be multiplied by the corresponding probability to obtain a total gray value, and then divided by the second pixel and the designated pixel To obtain an average gray value as the gray value of the second pixel containing the target noise.
- the designated pixel may be the second pixel in the image to be denoised.
- the pixels in the same row, or the pixels in the same column as the second pixel in the image to be denoised can also include the pixels in the same row and the second pixel in the image to be denoised.
- the pixels of the column By determining the total gray value of each row or column containing the target noise, and then dividing by the total number of pixels in the row or column, an average value is obtained, as each pixel in each row or column contains the target noise The gray value.
- the collected image to be denoised is an image of 4 ⁇ 4 pixels.
- the reference frequency band can determine that the frequencies corresponding to the four pixels in the first column are within the reference frequency band.
- the gray values of the four pixels in the first column are 6, 10, 12, and 5 in order.
- the reference gray value in the calibration data is 10, and the reference variance is 2.
- the probability of each pixel containing the target noise can be determined as 60%, 100%, 80%, 40%, and then the column containing the target noise can be calculated.
- the image to be denoised can be denoised according to the gray value of each pixel containing the target noise.
- the gray value of each pixel of the image to be denoised can be subtracted from the gray value of each pixel containing the target noise, that is, the denoised image can be obtained.
- a predetermined correction factor may be used to further correct the denoised image to obtain the final image. Among them, the correction factor can be determined based on empirical values.
- the noise of a fixed pattern will slowly change with temperature, time, etc.
- you can combine multiple frames of continuously acquired images to be denoised The gray value of the target noise is used to determine the gray value of each pixel in the image to be denoised in the current frame that contains the target noise.
- N frames of images before the image to be denoised in the current frame can be obtained, where N is a positive integer, and the specific value can be flexibly set according to the actual scene.
- an average gray value can be determined according to the gray value of each pixel of the image to be denoised in the current frame containing the target noise and the gray value of each pixel of the acquired first N frames of image containing the target noise, and the average gray value is taken as Each pixel of the image to be denoised in the current frame contains the gray value of the target noise.
- the average gray value may be an average value obtained by using each pixel on the image to be denoised in each frame as the granularity, or may be an average value obtained by using each row or each column of the image to be denoised in each frame as the granularity.
- the corresponding pixels of each pixel of the image to be denoised in the current frame in the N frames of image and the gray value of the target noise contained in these corresponding pixels can be determined respectively, and then each pixel can be calculated.
- the average value of the gray value of the target noise contained in the point and the gray value of the target noise contained in the corresponding pixel point of each pixel is referred to as the first average value, and then the calculated first average value is used as each pixel
- the point contains the average gray value of the target noise. That is, by averaging the target noise contained in the pixels representing the same three-dimensional object in each frame of image, the average value of the noise contained in the pixel in the current frame is obtained.
- the corresponding row of the row of each pixel of the current frame on the N frame image it is also possible to determine one by one the corresponding row of the row of each pixel of the current frame on the N frame image, and then calculate the gray value of the row of the pixel containing the target noise and the corresponding row.
- the corresponding row contains the average value of the gray value of the target noise, hereinafter referred to as the second average value, and then the second average value is divided by the number of pixels in the row, that is, each pixel in the row contains the target The average gray value of the noise.
- the value and the average value of the gray value of the corresponding column containing the target noise is called the third average value, and then dividing the third average value by the number of pixels in the column to get the average gray value of each pixel in the column containing the target noise value. That is, the target noise contained in the pixels in the same row or in the same column representing the same three-dimensional object in each frame of image can be averaged to obtain the average value of the target noise contained in each pixel in the row of the current frame.
- the calibration data can effectively distinguish real objects and fixed pattern noise, accurately estimate the gray value of fixed pattern noise, reduce the probability of misjudgment, and can reduce or eliminate the noise generated when the image is denoised. Artificial flaws, so as to achieve a better denoising effect.
- the calibration data is determined by referring to the image, and there is no need to reserve the masked area in the image sensor as a reference, which also improves the utilization rate of the image sensor and simplifies the difficulty of the manufacturing process of the image sensor.
- a specific embodiment is used to describe in detail below.
- the infrared image collected by the infrared sensor has fixed pattern noise vertical stripes, etc.
- the position of these vertical stripes in the image is relatively fixed. If there is no denoising process, there will be vertical stripes visible to the human eye on the infrared image. Affect the picture quality and temperature judgment.
- the current denoising method is used to remove the vertical stripes, there is a defect that the vertical stripes and the edges of the real object cannot be effectively identified, resulting in unsatisfactory denoising effects.
- an image processing method which mainly includes the following steps:
- the infrared sensor uses the infrared sensor to collect an image of a plane object as a reference image, and then perform Fourier transform on the image data of each column of the reference image to obtain the spectrogram corresponding to each column of image data, and then determine the vertical image according to the spectrogram.
- the frequency band where the fringe noise is located is used as the reference frequency band (assumed to be greater than 10KHZ), the average value of the gray value of each column containing noise is used as the reference gray value (assumed to be 10), and the variance of the gray value of each column is used as the reference variance (assuming Is 2).
- the reference frequency band, the reference gray value and the reference variance are stored in the designated location as calibration data.
- the original image collected by the infrared sensor without contrast stretching (assuming the image is a 4 ⁇ 4 image), and then perform Fourier transform on the number of images in each column of the original image to obtain the spectrogram of each column of image data. Then, according to the spectrogram and the predetermined reference frequency band of vertical fringe noise, determine whether the frequency corresponding to each column of the pixel in the image is within the reference frequency band. If the frequency of the pixel in this column is outside the reference frequency band (for example, If the frequency is less than 10KHZ), it is considered that this column does not contain vertical stripes, so the gray value of each pixel in this column containing vertical stripes noise is 0.
- the frequency of the pixels in this column is within the reference frequency band (for example, the frequency is greater than 10KHZ), it is considered that this column contains vertical stripes. Assume that the pixel frequencies of columns 1 and 3 are outside the reference frequency band, and the pixel frequencies of columns 2 and 4 are within the reference frequency band. The gray value of each pixel in the second column and the fourth column can be obtained.
- the gray value of each pixel in the second column is 6, 10, 12, and 5 respectively
- the gray value of each pixel is compared with the reference
- the present application also provides an image processing device.
- the device 30 includes a processor 31, a memory 32, and a computer program stored on the memory, and the processor executes the computer program.
- the calibration data includes a reference frequency band of the target noise, a reference gray value of the target noise, and a reference variance of the gray value of the target noise.
- the method when the processor is configured to determine the gray value of the target noise of each pixel of the image to be denoised according to the predetermined calibration data, the method includes:
- the gray value of the target noise of the first pixel is set to 0.
- the method when the processor is configured to determine the gray value of the target noise of each pixel of the image to be denoised according to the predetermined calibration data, the method includes:
- the gray value of the target noise contained in the second pixel is determined based on the gray value of the second pixel, the reference gray value, and the reference variance.
- the processor before the processor is configured to determine the gray value of the target noise contained in each pixel of the image to be denoised according to predetermined calibration data, it is further configured to:
- the pre-designed filter is used to perform filtering processing on the image to be denoised.
- the processor is configured to determine the gray value of the target noise of the second pixel based on the gray value of the second pixel, the reference gray value, and the reference variance. ,include:
- the gray value of the target noise of the second pixel is determined based on the gray value of the second pixel and the probability.
- the processor is configured to determine the probability that the second pixel includes the target noise based on the gray value of the second pixel, the reference gray value, and the reference variance ,include:
- the probability that the second pixel point includes the target noise is determined based on the distance and the reference variance.
- the distance includes one or more of difference, norm distance, Euclidean distance, Manhattan distance, Hamming distance, or cosine distance.
- the method when the processor is configured to determine the gray value of the target noise of the second pixel based on the gray value of the second pixel and the probability, the method includes:
- the average gray value is used as the gray value of the target noise contained in the second pixel.
- the designated pixels include:
- Pixels in the image to be denoised that are located in the same row as the second pixel;
- the method when the processor is configured to perform denoising processing on the image to be denoised according to the gray value of the target noise contained in each pixel, the method includes:
- N is a positive integer
- each pixel of the image to be denoised contains the average gray value of the target noise, and the average gray value is based on the gray value of each pixel of the image to be denoised containing the target noise and the N Determination of the gray value of each pixel of the frame image including the target noise;
- the method when the processor is configured to determine that each pixel of the image to be denoised contains the average gray value of the target noise, the method includes:
- the first average value is taken as the average gray value of each pixel including the target noise.
- the method when the processor is configured to determine that each pixel of the image to be denoised contains the average gray value of the target noise, the method includes:
- the reference image is a captured image of a planar object.
- the target noise includes horizontal stripe noise and/or vertical stripe noise.
- the image is an infrared image.
- the device 30 in addition to a processor 31, a memory 32, and a computer program stored on the memory, the device 30 also includes an infrared sensor 33, and the image to be denoised Obtained by the infrared sensor.
- the device is used in an unmanned aerial vehicle, and an infrared sensor is installed on the unmanned aerial vehicle to collect infrared images.
- the device can also be used in infrared thermal imaging cameras, infrared thermometers, or some aerial sounders and other products that use electromagnetic wave imaging, or used in some special image denoising processing.
- the devices such as laptops, mobile phones or cloud servers.
- an embodiment of the present specification also provides a computer storage medium in which a program is stored, and the program is executed by a processor to implement the image processing method in any of the foregoing embodiments.
- the embodiments of this specification may adopt the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program codes.
- Computer usable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be realized by any method or technology.
- the information can be computer-readable instructions, data structures, program modules, or other data.
- Examples of computer storage media include, but are not limited to: phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, CD-ROM, digital versatile disc (DVD) or other optical storage, Magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media can be used to store information that can be accessed by computing devices.
- PRAM phase change memory
- SRAM static random access memory
- DRAM dynamic random access memory
- RAM random access memory
- ROM read-only memory
- EEPROM electrically erasable programmable read-only memory
- flash memory or other memory technology
- CD-ROM compact disc
- DVD digital versatile disc
- Magnetic cassettes magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media can be used to store information that can be accessed by computing devices.
- the relevant part can refer to the part of the description of the method embodiment.
- the device embodiments described above are merely illustrative, where the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the objectives of the solutions of the embodiments. Those of ordinary skill in the art can understand and implement without creative work.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
一种图像处理方法、装置和存储介质。所述方法包括:获取待去噪图像;根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值;其中,所述标定数据基于参考图像中的所述目标噪声得到,用于确定所述目标噪声的频段和所述目标噪声的灰度值;根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理。通过标定数据作为参考,可以有效的从待去噪图像中识别噪声和真实物体,并且可以准确的估计噪声的灰度值,取得更好的去噪效果。
Description
本申请涉及图像处理技术领域,具体而言,涉及一种图像处理方法、装置及存储介质。
由于图像传感器的材料以及制造工艺的问题,图像传感器采集的图像中通常会包含一些固定模式的噪声,这些噪声固定的出现在图像传感器采集的每张图像上的固定位置。以红外传感器为例,由于制造工艺的限制,红外焦平面阵列上的各探测单元的响应特性不一致,各探测单元存在非均匀性,导致最终采集的图像会出现一些固定模式的噪声。噪声的存在会严重影响图像的清晰度和显示效果,因而需要对图像进行去噪处理。相关技术中在对图像进行去噪时,不能有效区别固定模式噪声和实际场景物体,导致去噪效果不理想,例如,不能有效判断竖条纹和竖直物体边缘,会导致竖直物体边缘上方产生竖直形态的人工瑕疵。因而,有必要对图像噪声去除的方法加以改进,提升图像的去噪效果。
发明内容
有鉴于此,本申请提供了一种图像处理方法、装置及存储介质。
根据本申请的第一方面,提供了一种图像处理方法,所述方法包括:
获取待去噪图像;
根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值;其中,所述标定数据基于参考图像中的所述目标噪声得到,用于确定所述目标噪声的频段和所述目标噪声的灰度值;
根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理。
根据本申请的第二方面,提供了一种图像处理装置,所述装置包括处理器、存储器以及存储在所述存储器上的计算机程序,所述处理器执行所述计算机程序时实现以下步骤:
获取待去噪图像;
根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值;其中,所述标定数据基于参考图像中的所述目标噪声得到,用于确定所述目标噪声的频段和所述目标噪声的灰度值;
根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理。
根据本申请的第三方面,提供一种计算机可读存储介质,所述计算机可读存储介质上存储有计算机程序,所述计算机程序被处理器执行时实现本申请任一项所述的图像处理方法。
应用本申请的方案,通过参考图像包含的目标噪声,预先确定用于确定目标噪声频段和灰度值的标定数据,然后根据标定确定待去噪图像中各像素点包含的目标噪声的灰度值,根据各像素点包含的目标噪声的灰度值对待去噪图像进行去噪处理。通过标定数据作为参考,可以有效的从待去噪图像中识别噪声和真实物体,并且可以准确的估计噪声的灰度值,取得更好的去噪效果,提升红外去噪的准确性。
为了更清楚地说明本申请实施例中的技术方案,下面将对实施例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是本发明一个实施例提供的一种包含竖条纹噪声的图像。
图2是本发明一个实施例提供的一种图像处理方法的流程图。
图3是本发明一个实施例提供的一种图像去噪装置的逻辑结构框图。
图4是本发明一个实施例提供的另一种图像去噪装置的逻辑结构框图。
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,所描述的实施例仅仅是本申请一部分实施例,而不是全部的实施例。
由于图像传感器的材料以及制造工艺的问题,图像传感器采集的图像中通常会包含一些固定模式的噪声,这些噪声固定的出现在图像传感器采集的每张图像上。以红外传感器为例,由于制造工艺的限制,红外焦平面阵列上的各探测单元的响应特性不一致,各探测单元存在非均匀性,导致最终采集的图像会出现一些固定模式的噪声。比如,如果红外焦平面阵列同一行的探测单元共享一个输出电路,由于各行输出电路偏置电压的差异,导致相邻两行的像素点的灰度值会出现明显的差异,最终采集的图像上会出现横条纹噪声,如果红外焦平面阵列同一列的探测单元共享一个输出电路,那么由于各列输出电路偏置电压的差异,就会导致图像上出现大量的竖条纹噪声。当然,如果红外焦平面阵列的各探测单元的输出电路是按其他模式共享或者一个探测单元独立采用一个输出电路,那么就会出现其他模式的固定噪声。如图1所示,采集得到的图像中包含很多条竖条纹噪声。
噪声的存在会严重影响图像的清晰度度和显示效果,尤其是对于红外图像这种分辨率本来就比较低的图像。因而,有必要对图像进行去噪处理。
相关技术在对图像进行去噪处理时,有的采用频域高通滤波法,即将图像数据从空间域转化到频域,由于噪声往往处于高频,因而可以过滤出高频部分,根据高频部分的像素点的灰度值得到噪声的灰度值,进而根据噪声的灰度值对图像进行去噪处理。但是这种方式有些时候无法有效的区别噪声和真实场景中的物体,尤其是当真实物体本身差异较大时,比如真 实物体是棱角较多的物体,这时就比较难分别噪声和真实物体的棱角,比如无法区分竖条纹和物体边缘。这情况下采用该方法对图像噪声进行去噪处理后会产生一些瑕疵,去噪效果不理想。
有些技术会在图像传感器中设置一块遮挡区域,这个遮挡区域的探测单元无法感光,因而,这块区域中各探测单元对应的输出值即体现了探测单元的响应差异,进而可以作为参考,确定因探测单元的响应差异产生的噪声,并对图像进行去噪处理。比如每一行或列保留一个或多个探测单元被遮挡,无法感光,从而该行或列的噪声可以根据被遮挡的探测单元的输出值确定。但是这种方式需要图像图传感器保留一块遮挡区域,浪费了传感器的空间,也增加了传感器的制造工艺难度。
基于此,本申请提供了一种图像处理方法,由于噪声为固定模式的噪声,因而可以通过参考图像上的噪声确定标定数据,该标定数据用于确定噪声的参考频段和参考灰度值,然后根据标定数据从待去噪图像中有效地识别出噪声,并进行去噪处理,从达到更好的去噪效果。具体的,如图2所示,所述方法可包括以下步骤;
S202、获取待去噪图像;
S204、根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值;其中,所述标定数据基于参考图像中的所述目标噪声得到,用于确定所述目标噪声的频段和所述目标噪声的灰度值;
S206、根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理。
本申请的图像处理方法可以用于各种图像采集装置中,比如,红外热像仪,图像采集装置在采集到图像后即可以直接进行去噪处理,当然也可以用于其他的对图像进行后处理的电子设备中,该电子设备从图像采集装置获取采集到待去噪图像,然后进行去噪处理。
本申请实施例中,像素点的灰度值可以指温度值。
本申请的待去噪图像可以是各种遥感图像,遥感图像为通过接收探测 目标物电磁辐射信息得到的图像,比如,在某些实施例中,该待去噪图像可以是通过红外传感器采集的红外图像,当然,待去噪图像也可以是通过其他电磁波得到的图像,本申请不作限制。
本申请的目标噪声为固定模式的噪声,由于该噪声产生的主要原因是探测单元输出电路存在偏差,因而针对同一个传感器采集的图像,该噪声在图像中出现的位置基本固定。固定模式的噪声包括各种条纹噪声,比如,在某些实施例中,目标噪声可以是横条纹噪声、竖条纹噪声或者既包含横条纹噪声,也包含竖条纹噪声。在某些实施例中,目标噪声也可以是暗影、鬼点等在图像固定位置出现的其他噪声。
由于噪声产生的主要原因是探测单元输出电路存在偏差,导致图像中某些像素点的灰度值相比于邻近的像素点突然偏高或偏低,因而噪声通常包含在灰度值急剧变化的像素点。所以,可以通过频域滤波方法将高频部分过滤出来,从而得到噪声。但是这种方法往往会把真实物体中变化较大的部分也当成噪声。为了有效地区别噪声和真实物体,可以通过一参考图像中包含的目标噪声,确定标定数据,通过该标定数据来确定待去噪图像中的目标噪声所在的频段和目标噪声的灰度值,使得确定的目标噪声的灰度值更加准确。
在某些实施例中,参考图像可以是通过图像传感器采集得到的一张平面物体的图像,其中,该图像传感器为采集待去噪图像的传感器。由于平面物体表面基本一致,不存在棱角等比较突兀的地方,因而采集得到的图像上相比于邻近像素点灰度值差异较大的像素点基本可以认为都是噪声引起的,因而图像上显示出来的灰度值急剧变化的部分都是噪声,而不是真实物体本身。因而,在采集到参考图像后,可以对参考图像进行傅里叶变换或者采用预先设计的高通滤波器对参考图像进行滤波处理,以得到该标定数据。
在某些实施中,标定数据可以包括目标噪声的参考频段、目标噪声的参考灰度值以及目标噪声灰度值的参考方差。其中,该参考频段、参考灰 度值和参考方差都是根据参考图像中的目标噪声得到的,用于后续对图像进行去噪时作为参考。当然,在某些实施例中,标定数据也可以只包括目标噪声的频段和灰度值。其中,目标噪声的参考灰度值可以是整张图像的噪声的灰度值的平均值,也可以是每一行或者每一列的噪声灰度值的平均值,具体可以根据实际需求去设定。
在得到参考图像后,可以对参考图像进行傅里叶变化,得到参考图像对应的频谱图,其中,可以对参考图像进行一维的傅里叶变换,比如对每一行或者每一列的图像数据进行傅里叶变化,得到每一行或每一列图像数据的频谱图,然后可以根据每一行或列的频谱图确定目标噪声所在的频段以及每一行或每一列的噪声的平均灰度值,作为所述参考灰度值,然后,根据每一行或每一列的平均灰度值计算得到一个灰度值的方差,作为所述参考方差。当然,也可以对参考图像的图像数据进行二维的傅里叶变换,得到整张图像的频谱图,然后根据频谱图确定目标噪声所在的频段以及整张图像的噪声灰度值的平均值。此外,还可以采用预先设计的高通滤波器对参考图像进行滤波处理,得到所述参考频段参考灰度值以及参考方差。具体采用哪种方式去根据参考图像确定标定数据,可以根据实际情况确定,本申请不作限制。
在确定标定数据后,可以获取图像传感器采集的待去噪图像。为了与参考图像保持一致,如果参考图像是未经过对比度拉伸的原始图像,那么待去噪图像也可以是传感器采集的未经过对比度拉伸的原始图像。然后可以根据预先确定的标定数据确定待去噪图像中的各个像素点包含目标噪声的灰度值,并根据各像素点包含噪声的灰度值对待去噪图像进行去噪处理。
为了确定待去噪图像中各像素点的目标噪声的灰度值,可以根据标定数据中目标噪声所在的参考频段从待去噪图像中确定包含目标噪声的像素点有哪些,不包含目标噪声的像素点有哪些,以及包含目标噪声的像素点中目标噪声的灰度值是多少。在某些实施例中为了确定待去噪图像中各像素点是否包含目标噪声,可以先对待去噪图像进行傅里叶变化得到待去噪 图像的频谱图,或者也可以通过预先设计的高通滤波器对待去噪图像进行滤波处理,然后再根据标定数据中的参考频段确定各像素点是否包含目标噪声。比如可以对待去噪图像中的每一行或者每一列的图像数据进行傅里叶变化,得到每一行或每一列图像数据的频谱图,然后可以根据每一行或每一列的频谱图以及标定数据中的参考频段,确定该行或者该列是否包含目标噪声。
在某些实施例中,在对待去噪图像进行傅里叶变化,或者采用高通滤波器进行过滤处理后,可以根据标定数据中的参考频段从待去噪图像中确定第一像素点,其中第一像素点为参考频段之外的像素点,对于参考频段之外的像素点,可以认为不包含目标噪声,因而,可以将这些像素点包含目标噪声的灰度值设为0。举个例子,根据参考图像确定出来目标噪声所在的参考频段为大于10KHZ,那么便可以认为频率大于10KHZ的像素点为包含目标噪声像素点,如果频率小于10KHZ,则认为这个像素点不包含目标噪声,因而这个像素点包含目标噪声的灰度值为0。
在某些实施例中,在对待去噪图像进行傅里叶变化,或者采用高通滤波器进行过滤处理后,可以根据标定数据中的参考频段从待去噪图像中确定第二像素点,其中第二像素点为参考频段之内的像素点,对于参考频段之内的像素点,可以认为这些像素点包含目标噪声,因而可以进一步根据标定数据中的参考灰度值以及参考方差确定出这些像素点中包含目标噪声的灰度值。
由于待去噪图像的中物体并非都是平面物体,因而待去噪图像中出现灰度值变化较大的像素点可能是噪声和真实物体不同部位存在差异综合的结果。为了更加准确地计算第二像素点包含目标噪声的灰度值,可以根据第二像素点的灰度值、标定数据中的参考灰度值以及参考方差先确定每个第二像素点包含目标噪声的概率,然后根据每个第二像素点包含目标噪声的概率以及第二像素点的灰度值确定各第二像素点包含目标噪声的灰度值。
其中,在某些实施例中,在确定每个第二像素点包含目标噪声的概率时,可以确定第二像素点的灰度值与参考灰度值的距离,然后根据该距离以及参考方差确定每个第二像素点包含目标噪声的概率。由于根据参考图像确定的参考灰度值可以是各行或各列包含目标噪声的灰度值的平均值,参考方差也是根据各行或各列方差的平均灰度值得到,因而说明目标噪声的灰度值可以在参考方差内波动,假设参考灰度值为10,参考方差为2,说明目标噪声的灰度值可以在8~12之间。因此,可以确定第二像素点的灰度值与参考灰度值的距离,然后再将该距离与方差比较,如果该距离与方差越接近,则说明该第二像素点包含目标噪声的概率越大,该距离与方差越小,则说明该第二像素点包含目标噪声的概率越小。
在某些实施例中,所述距离可以是差值、范数距离、欧式距离、曼哈顿距离、汉明距离或余弦距离的一种或多种。其中,范数距离可以是二范数距离,或者其他各种范数空间的距离。比如,目标噪声的参考灰度值为10,而第二像素点的灰度值为8,因此,可以算这两个灰度值之间的上述一种或者多种距离,然后再跟方差比较,以此来确定各个第二像素点包含目标噪声的概率。
在确定各个第二像素点包含目标噪声的概率后,可以根据第二像素点的灰度值以及第二像素点包含目标噪声的概率确定第二像素点包含目标噪声的灰度值。在某些实施例中,可以根据第二像素点以及指定像素点的灰度值以及包含目标噪声的概率来确定第二像素点包含目标噪声的灰度值。比如可以用第二像素点的灰度值乘以对应的概率,以及指定像素点的灰度值乘以对应的概率,得到一个灰度值总量,再除以第二像素点和指定像素点的总数量,得到一个平均灰度值,作为所述第二像素点包含目标噪声的灰度值。
由于目标噪声通常为横条纹噪声或竖条纹噪声,即在图像固定的行或固定的列出现,因而在某些实施例中,指定像素点可以是该待去噪图像中与第二像素点位于同一行的像素点,或者是该待去噪图像中与第二像素点 位于同一列的像素点,也可以同时包括该待去噪图像中与第二像素点位于同一行的像素点和位于同一列的像素点。通过确定每一行或每一列包含目标噪声的灰度值的总量,然后除以该行或列的像素点的总数量,得到一个平均值,作为每一行或每一列中各像素点包含目标噪声的灰度值。举个例子,假设采集得到的待去噪图像为4×4个像素点的图像,对该待去噪图像的每一列进行傅里叶变换,得到各列的频谱图,然后通过标定数据中的参考频段可以确定第一列的四个像素点对应的频率在参考频段内。假设第一列的四个像素点灰度值依次为6、10、12、5。标定数据中的参考灰度值为10,参考方差为2。根据四个像素点的灰度值,参考灰度值以及参考方差可以确定各像素点包含目标噪声的概率分别为60%、100%、80%、40%,然后可以计算这一列包含目标噪声的灰度值的总量:6×60%+10×100%+12×80%+5×40%=25.2,则该列每一个像素点包含目标噪声的灰度值的平均值为:25.2÷4=6.3。然后可以根据这个灰度值对这一列的像素点进行去噪处理。
在确定各像素点包含目标噪声的灰度值后,可以根据各像素点包含目标噪声的灰度值对待去噪图像进行去噪处理。在某些实施例中,可以将待去噪图像各像素点的灰度值减去各像素点包含目标噪声的灰度值,即可以得到去噪后的图像。当然,在某些实施例中,还可以采用预先确定的校正因子对去噪后的图像做进一步地校正,得到最终的图像。其中,校正因子可以根据经验值确定。
当然,在某些场景中,固定模式的噪声会随温度、时间等发生缓慢变化,为了更加准确地确定出各像素点包含目标噪声的灰度值,可以结合多帧连续获取的待去噪图像中目标噪声的灰度值来确定当前帧待去噪图像中的各像素点包含目标噪声的灰度值。比如,在某些实施例中,可以获取当前帧待去噪图像之前的N帧图像,其中,N为正整数,具体数值可以根据实际场景灵活设置。然后可以根据当前帧待去噪图像各像素点包含目标噪声的灰度值和获取的前N帧图像各像素点包含目标噪声的灰度值确定一个 平均灰度值,将该平均灰度值作为当前帧待去噪图像各像素点包含目标噪声的灰度值。
其中,该平均灰度值可以是以各帧待去噪图像上的各像素点为粒度得到一个均值,也可以是以各帧待去噪图像上的各行或各列为粒度得到的一个均值。比如,在某些实施例中,可以分别确定当前帧待去噪图像各像素点在该N帧图像的对应像素点,以及这些对应像素点包含的目标噪声的灰度值,然后可以计算各像素点包含的目标噪声的灰度值与各像素点的对应像素点包含的目标噪声的灰度值的平均值,以下称为第一平均值,然后将计算得到的该第一平均值作为各像素点包含目标噪声的平均灰度值。即通过对各帧图像中表示同一个三维物体的像素点包含的目标噪声取平均,得到当前帧的该像素点包含噪声的平均值。
在某些实施例中,也可以逐一确定当前帧的各像素点所在的行在该N帧图像上的对应行,然后计算所述像素点所在的行包含所述目标噪声的灰度值与所述对应行包含所述目标噪声的灰度值的平均值,以下称为第二平均值,然后将第二平均值除以该行像素点的个数,即可以得到该行各像素点包含目标噪声的平均灰度值。当然,在某些实施例中,也可以先逐一确定当前帧待去噪图像各像素点所在的列在该N帧图像上的对应列,然后计算该像素点所在的列包含目标噪声的灰度值与对应列包含目标噪声的灰度值的平均值,称为第三平均值,再用该第三平均值除以该列像素点的数量得到该列各像素点包含目标噪声的平均灰度值。即可以通过对各帧图像中表示同一个三维物体的同一行的像素点或同一列的像素点包含的目标噪声取平均,以得到当前帧的该行的各像素点包含目标噪声的平均值。
通过本申请提供的图像处理方法,可以通过标定数据有效区别真实物体和固定模式噪声,准确估计固定模式噪声的灰度值,降低误判概率,并且可以降低或消除对图像进行去噪处理时产生的人工瑕疵,从而取得更好的去噪效果。同时,通过参考图像确定标定数据,无需在图像传感器保留遮挡区域作为参考,也提高了图像传感器的利用率,简化了图像传感器的 制造工艺的难度。为了进一步解释本申请提供的图像处理方法,以下以一个具体实施例详细说明。
由于生产制造问题,红外传感器采集的红外图像带有固定模式噪声竖条纹等,这些竖条纹在图像中的位置相对固定,如果不经去噪处理,红外图像上会有人眼可见的竖条纹,严重影响画质和温度判断。采用目前的去噪方法去除竖条纹时,存在一缺陷,即无法有效识别竖条纹和真实物体的边缘,导致去噪效果不理想。
为了有效的去除红外图像上的竖条纹噪声,提出一种图像处理方法,主要包括以下步骤:
(1)标定数据的确定
使用该红外传感器采集一张平面物体的图像作为参考图像,然后对该参考图像的每一列的图像数据进行傅里叶变换,得到每一列图像数据对应的频谱图,然后根据频谱图确定出该竖条纹噪声所在的频段作为参考频段(假设为大于10KHZ)、各列包含噪声的灰度值的平均值作为参考灰度值(假设为10),以及各列灰度值的方差作为参考方差(假设为2)。其中,将参考频段、参考灰度值和参考方差作为标定数据存储在指定位置。
(2)对红外传感器采集的图像进行去噪
获取红外传感器采集的未经过对比度拉伸的原始图像(假设该图像为4×4图像),然后对该原始图像的每一列图像数进行傅里叶变换,得到每一列图像数据的频谱图。然后根据该频谱图和预先确定的竖条纹噪声的参考频段确定该图像中每一列的像素点对应的频率是否在该参考频段之内,如果这一列的像素点的频率在参考频段之外(比如频率小于10KHZ),则认为这一列不包含竖条纹,因而这一列每个像素点的包含竖条纹噪声的灰度值为0。如果这一列的像素点的频率在参考频段之内(比如频率大于10KHZ),则认为这一列包含竖条纹。假设1,3列的像素点频率在参考频段之外,2,4列的像素点频率在参考频段之内。可以获取第2列和第4列各像素点的灰度值,假设第2列的各像素点会灰度值分别为6、10、12、5, 然后将各像素点的灰度值与参考灰度值(10)比较,计算各像素点灰度值与参考灰度值的余弦距离,然后根据余弦距离和参考方差的接近成都确定各像素点包含竖条纹噪声的概率。假设分别为60%、100%、80%、40%,然后可以计算第2列包含竖条纹噪声的灰度值的总量:6×60%+10×100%+12×80%+5×40%=25.2。然后可以获取该图像之前的3帧图像,确定第2列在这三帧图像对应的列,然后确定对应的列包含竖条纹噪声的灰度值的总量分别为26、24.8、24,因而可以求得各帧图像该列的包含竖条纹噪声灰度值的平均值:(25.2+26+24.8+24)/4=25,从而可以确定第2列每个像素点包含竖条纹噪声的灰度值为25/4=6.25。针对第4列,可以用同样的方法求得各像素点包含竖条纹噪声的灰度值,假设为7,。然后将第1列和第3列各像素点的灰度值减去0,第2列的各像素点的灰度值减去6.25,第4列的各像素点的灰度值减去7,从而得到去噪后的图像。
通过这种方法,可以有效区别真实物体和固定模式噪声,准确估计固定模式噪声的灰度值,降低误判概率,并且可以降低或消除对图像进行去噪处理时产生的人工瑕疵,从而使红外图像更干净,温度判断更准确。
另外,本申请还提供了一种图像处理装置,如图3所示,所述装置30包括处理器31、存储器32以及存储在所述存储器上的计算机程序,所述处理器执行所述计算机程序时实现以下步骤:
获取待去噪图像;
根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值;其中,所述标定数据基于参考图像中的所述目标噪声得到,用于确定所述目标噪声的频段和所述目标噪声的灰度值;
根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理。
在某些实施例中,所述标定数据包括所述目标噪声的参考频段、所述目标噪声的参考灰度值以及所述目标噪声灰度值的参考方差。
在某些实施例中,所述处理器用于根据预先确定的标定数据确定所述 待去噪图像各像素点的目标噪声的灰度值时,包括:
基于所述参考频段确定所述待去噪图像中的第一像素点,所述第一像素点为所述参考频段之外的像素点;
将所述第一像素点的目标噪声的灰度值设为0。
在某些实施例中,所述处理器用于根据预先确定的标定数据确定所述待去噪图像各像素点的目标噪声的灰度值时,包括:
基于所述参考频段确定所述待去噪图像中的第二像素点,所述第二像素点为所述参考频段之内的像素点;
基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包含的目标噪声的灰度值。
在某些实施例中,所述处理器用于根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值之前,还用于:
对所述待去噪图像进行傅里叶变换,得到所述待去噪图像频谱图;或
采用预先设计的滤波器对所述待去噪图像进行滤波处理。
在某些实施例中,所述处理器用于基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点的目标噪声的灰度值时,包括:
基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包括所述目标噪声的概率;
基于所述第二像素点的灰度值和所述概率确定所述第二像素点的目标噪声的灰度值。
在某些实施例中,所述处理器用于基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包括所述目标噪声的概率时,包括:
确定所述第二像素点的灰度值与所述参考灰度值之间的距离;
基于所述距离以及所述参考方差确定所述第二像素点包括所述目标噪声的概率。
在某些实施例中,所述距离包括:差值、范数距离、欧式距离、曼哈顿距离、汉明距离或余弦距离的一种或多种。
在某些实施例中,所述处理器用于基于所述第二像素点的灰度值和所述概率确定所述第二像素点的目标噪声的灰度值时,包括:
基于所述第二像素点的灰度值、所述第二像素点对应的所述概率、指定像素点的灰度值以及所述指定像素点对应的所述概率,确定所述第二像素点和所述指定像素点包含的所述目标噪声的平均灰度值;
将所述平均灰度值作为所述第二像素点包含的目标噪声的灰度值。
在某些实施例中,所述指定像素点包括:
所述待去噪图像中与所述第二像素点位于同一行的像素点;和/或
所述待去噪图像中与所述第二像素点位于同一列的像素点。
在某些实施例中,所述处理器用于根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理时,包括:
获取所述待去噪图像之前的N帧图像,N为正整数;
确定所述待去噪图像各像素点包含所述目标噪声的平均灰度值,所述平均灰度值基于所述待去噪图像各像素点包含所述目标噪声的灰度值和所述N帧图像各像素点包含所述目标噪声的灰度值确定;
基于所述平均灰度值对所述待去噪图像进行去噪处理。
在某些实施例中,所述处理器用于确定所述待去噪图像各像素点包含所述目标噪声的平均灰度值时,包括:
分别确定所述待去噪图像各像素点在所述N帧图像的对应像素点,以及所述对应像素点包含的所述目标噪声的灰度值;
计算各像素点包含的目标噪声的灰度值与各像素点的对应像素点包含的所述目标噪声的灰度值的第一平均值;
将所述第一平均值作为所述各像素点包含所述目标噪声的平均灰度值。
在某些实施例中,所述处理器用于确定所述待去噪图像各像素点包含 所述目标噪声的平均灰度值时,包括:
确定所述各像素点所在的行在所述N帧图像上的对应行;
计算所述像素点所在的行包含所述目标噪声的灰度值与所述对应行包含目标噪声的灰度值的第二平均值;
基于所述第二平均值确定所述各像素点包含所述目标噪声的平均灰度值;和/或
计算所述各像素点所在的列在所述N帧图像上的对应列;
确定所述像素点所在的列包含所述目标噪声的灰度值与所述对应列包含目标噪声的灰度值的第三平均值;
基于所述第三平均值确定所述各像素点包含所述目标噪声的平均灰度值。
在某些实施例中,所述参考图像为采集的平面物体的图像。
在某些实施例中,所述目标噪声包括:横条纹噪声和/或竖条纹噪声。
在某些实施例中,所述图像为红外图像。
在某些实施例中,如图4所示,所述装置30除了包括处理器31、存储器32以及存储在所述存储器上的计算机程序之外,还包括红外传感器33,所述待去噪图像通过所述红外传感器采集得到。
在某些实施例中,所述装置用于无人机,无人机上安装有红外传感器,用于采集红外图像。
当然,在某些实施例中,所述装置也可以用于红外热像仪、红外测温仪或者是一些航空探测仪等采用电磁波成像的产品当中,或者用于一些专门对图像进行去噪处理的设备当中,比如笔记本电脑、手机或者云端服务器。
相应地,本说明书实施例还提供一种计算机存储介质,所述存储介质中存储有程序,所述程序被处理器执行时实现上述任一实施例中图像处理方法。
本说明书实施例可采用在一个或多个其中包含有程序代码的存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程 序产品的形式。计算机可用存储介质包括永久性和非永久性、可移动和非可移动媒体,可以由任何方法或技术来实现信息存储。信息可以是计算机可读指令、数据结构、程序的模块或其他数据。计算机的存储介质的例子包括但不限于:相变内存(PRAM)、静态随机存取存储器(SRAM)、动态随机存取存储器(DRAM)、其他类型的随机存取存储器(RAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、快闪记忆体或其他内存技术、只读光盘只读存储器(CD-ROM)、数字多功能光盘(DVD)或其他光学存储、磁盒式磁带,磁带磁磁盘存储或其他磁性存储设备或任何其他非传输介质,可用于存储可以被计算设备访问的信息。
对于装置实施例而言,由于其基本对应于方法实施例,所以相关之处参见方法实施例的部分说明即可。以上所描述的装置实施例仅仅是示意性的,其中所述作为分离部件说明的单元可以是或者也可以不是物理上分开的,作为单元显示的部件可以是或者也可以不是物理单元,即可以位于一个地方,或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。本领域普通技术人员在不付出创造性劳动的情况下,即可以理解并实施。
需要说明的是,在本文中,诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来,而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含,从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素,而且还包括没有明确列出的其他要素,或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下,由语句“包括一个……”限定的要素,并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
以上对本发明实施例所提供的方法和装置进行了详细介绍,本文中应用了具体个例对本发明的原理及实施方式进行了阐述,以上实施例的说明 只是用于帮助理解本发明的方法及其核心思想;同时,对于本领域的一般技术人员,依据本发明的思想,在具体实施方式及应用范围上均会有改变之处,综上所述,本说明书内容不应理解为对本发明的限制。
Claims (35)
- 一种图像处理方法,其特征在于,所述方法包括:获取待去噪图像;根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值;其中,所述标定数据基于参考图像中的所述目标噪声得到,用于确定所述目标噪声的频段和所述目标噪声的灰度值;根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理。
- 根据权利要求1所述的图像处理方法,其特征在于,所述标定数据包括所述目标噪声的参考频段、所述目标噪声的参考灰度值以及所述目标噪声灰度值的参考方差。
- 根据权利要求2所述的图像处理方法,其特征在于,根据预先确定的标定数据确定所述待去噪图像各像素点的目标噪声的灰度值,包括:基于所述参考频段确定所述待去噪图像中的第一像素点,所述第一像素点为所述参考频段之外的像素点;将所述第一像素点的目标噪声的灰度值设为0。
- 根据权利要求2所述的图像处理方法,其特征在于,所述根据预先确定的标定数据确定所述待去噪图像各像素点的目标噪声的灰度值,包括:基于所述参考频段确定所述待去噪图像中的第二像素点,所述第二像素点为所述参考频段之内的像素点;基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包含的目标噪声的灰度值。
- 根据权利要求1至4任一项所述的图像处理方法,所述根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值之前,还包括:对所述待去噪图像进行傅里叶变换,得到所述待去噪图像频谱图;或采用预先设计的滤波器对所述待去噪图像进行滤波处理。
- 根据权利要求4所述的图像处理方法,其特征在于,基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点的目标噪声的灰度值,包括:基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包括所述目标噪声的概率;基于所述第二像素点的灰度值和所述概率确定所述第二像素点的目标噪声的灰度值。
- 根据权利要求6所述的图像处理方法,其特征在于,基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包括所述目标噪声的概率,包括:确定所述第二像素点的灰度值与所述参考灰度值之间的距离;基于所述距离以及所述参考方差确定所述第二像素点包括所述目标噪声的概率。
- 根据权利要求7所述的图像处理方法,其特征在于,所述距离包括:差值、范数距离、欧式距离、曼哈顿距离、汉明距离或余弦距离的一种或多种。
- 根据权利要求6所述的图像处理方法,其特征在于,基于所述第二像素点的灰度值和所述概率确定所述第二像素点的目标噪声的灰度值,包括:基于所述第二像素点的灰度值、所述第二像素点对应的所述概率、指定像素点的灰度值以及所述指定像素点对应的所述概率,确定所述第二像素点和所述指定像素点包含的所述目标噪声的平均灰度值;将所述平均灰度值作为所述第二像素点包含的目标噪声的灰度值。
- 根据权利要求9所述的图像处理方法,其特征在于,所述指定像素点包括:所述待去噪图像中与所述第二像素点位于同一行的像素点;和/或所述待去噪图像中与所述第二像素点位于同一列的像素点。
- 根据权利要求1所述的图像处理方法,其特征在于,根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理,包括:获取所述待去噪图像之前的N帧图像,N为正整数;确定所述待去噪图像各像素点包含所述目标噪声的平均灰度值,所述平均灰度值基于所述待去噪图像各像素点包含所述目标噪声的灰度值和所述N帧图像各像素点包含所述目标噪声的灰度值确定;基于所述平均灰度值对所述待去噪图像进行去噪处理。
- 根据权利要求11所述的图像处理方法,其特征在于,确定所述待去噪图像各像素点包含所述目标噪声的平均灰度值,包括:分别确定所述待去噪图像各像素点在所述N帧图像的对应像素点,以及所述对应像素点包含的所述目标噪声的灰度值;计算各像素点包含的目标噪声的灰度值与各像素点的对应像素点包含的所述目标噪声的灰度值的第一平均值;将所述第一平均值作为所述各像素点包含所述目标噪声的平均灰度值。
- 根据权利要求11所述的图像处理方法,其特征在于,确定所述待去噪图像各像素点包含所述目标噪声的平均灰度值,包括:确定所述各像素点所在的行在所述N帧图像上的对应行;计算所述像素点所在的行包含所述目标噪声的灰度值与所述对应行包含所述目标噪声的灰度值的第二平均值;基于所述第二平均值确定所述各像素点包含所述目标噪声的平均灰度值;和/或确定所述各像素点所在的列在所述N帧图像上的对应列;计算所述像素点所在的列包含所述目标噪声的灰度值与所述对应列包含目标噪声的灰度值的第三平均值;基于所述第三平均值确定所述各像素点包含所述目标噪声的平均灰度值。
- 根据权利要求1-13任一项所述的图像处理方法,其特征在于,所述参考图像为采集的平面物体的图像。
- 根据权利要求1-14任一项所述的图像处理方法,其特征在于,所述目标噪声包括:横条纹噪声和/或竖条纹噪声。
- 根据权利要求1-15任一项所述的图像处理方法,其特征在于,所述图像为红外图像。
- 一种图像处理装置,其特征在于,所述装置包括处理器、存储器以及存储在所述存储器上的计算机程序,所述处理器执行所述计算机程序时实现以下步骤:获取待去噪图像;根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值;其中,所述标定数据基于参考图像中的所述目标噪声得到,用于确定所述目标噪声的频段和所述目标噪声的灰度值;根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理。
- 根据权利要求17所述的图像处理装置,其特征在于,所述标定数据包括所述目标噪声的参考频段、所述目标噪声的参考灰度值以及所述目标噪声灰度值的参考方差。
- 根据权利要求18所述的图像处理装置,其特征在于,所述处理器用于根据预先确定的标定数据确定所述待去噪图像各像素点的目标噪声的灰度值时,包括:基于所述参考频段确定所述待去噪图像中的第一像素点,所述第一像素点为所述参考频段之外的像素点;将所述第一像素点的目标噪声的灰度值设为0。
- 根据权利要求18所述的图像处理装置,其特征在于,所述处理器用于根据预先确定的标定数据确定所述待去噪图像各像素点的目标噪声的灰度值时,包括:基于所述参考频段确定所述待去噪图像中的第二像素点,所述第二像素点为所述参考频段之内的像素点;基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包含的目标噪声的灰度值。
- 根据权利要求17-20任一项所述的图像处理装置,所述处理器用于根据预先确定的标定数据确定所述待去噪图像各像素点包含的目标噪声的灰度值之前,还用于:对所述待去噪图像进行傅里叶变换,得到所述待去噪图像频谱图;或采用预先设计的滤波器对所述待去噪图像进行滤波处理。
- 根据权利要求20所述的图像处理装置,其特征在于,所述处理器用于基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点的目标噪声的灰度值时,包括:基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包括所述目标噪声的概率;基于所述第二像素点的灰度值和所述概率确定所述第二像素点的目标噪声的灰度值。
- 根据权利要求22所述的图像处理装置,其特征在于,所述处理器用于基于所述第二像素点的灰度值、所述参考灰度值以及所述参考方差确定所述第二像素点包括所述目标噪声的概率时,包括:确定所述第二像素点的灰度值与所述参考灰度值之间的距离;基于所述距离以及所述参考方差确定所述第二像素点包括所述目标噪声的概率。
- 根据权利要求23所述的图像处理装置,其特征在于,所述距离包括:差值、范数距离、欧式距离、曼哈顿距离、汉明距离或余弦距离的一种或多种。
- 根据权利要求22所述的图像处理装置,其特征在于,所述处理器用于基于所述第二像素点的灰度值和所述概率确定所述第二像素点的目标 噪声的灰度值时,包括:基于所述第二像素点的灰度值、所述第二像素点对应的所述概率、指定像素点的灰度值以及所述指定像素点对应的所述概率,确定所述第二像素点和所述指定像素点包含的所述目标噪声的平均灰度值;将所述平均灰度值作为所述第二像素点包含的目标噪声的灰度值。
- 根据权利要求25所述的图像处理装置,其特征在于,所述指定像素点包括:所述待去噪图像中与所述第二像素点位于同一行的像素点;和/或所述待去噪图像中与所述第二像素点位于同一列的像素点。
- 根据权利要求17所述的图像处理装置,其特征在于,所述处理器用于根据各像素点包含的目标噪声的灰度值对所述待去噪图像进行去噪处理时,包括:获取所述待去噪图像之前的N帧图像,N为正整数;确定所述待去噪图像各像素点包含所述目标噪声的平均灰度值,所述平均灰度值基于所述待去噪图像各像素点包含所述目标噪声的灰度值和所述N帧图像各像素点包含所述目标噪声的灰度值确定;基于所述平均灰度值对所述待去噪图像进行去噪处理。
- 根据权利要求27所述的图像处理装置,其特征在于,所述处理器用于确定所述待去噪图像各像素点包含所述目标噪声的平均灰度值时,包括:分别确定所述待去噪图像各像素点在所述N帧图像的对应像素点,以及所述对应像素点包含的所述目标噪声的灰度值;计算各像素点包含的目标噪声的灰度值与各像素点的对应像素点包含的所述目标噪声的灰度值的第一平均值;将所述第一平均值作为所述各像素点包含所述目标噪声的平均灰度值。
- 根据权利要求27所述的图像处理装置,其特征在于,所述处理器 用于确定所述待去噪图像各像素点包含所述目标噪声的平均灰度值时,包括:确定所述各像素点所在的行在所述N帧图像上的对应行;计算所述像素点所在的行包含所述目标噪声的灰度值与所述对应行包含目标噪声的灰度值的第二平均值;基于所述第二平均值确定所述各像素点包含所述目标噪声的平均灰度值;和/或计算所述各像素点所在的列在所述N帧图像上的对应列;确定所述像素点所在的列包含所述目标噪声的灰度值与所述对应列包含目标噪声的灰度值的第三平均值;基于所述第三平均值确定所述各像素点包含所述目标噪声的平均灰度值。
- 根据权利要求17-29任一项所述的图像处理装置,其特征在于,所述参考图像为采集的平面物体的图像。
- 根据权利要求17-30任一项所述的图像处理装置,其特征在于,所述目标噪声包括:横条纹噪声和/或竖条纹噪声。
- 根据权利要求17-31任一项所述的图像处理装置,其特征在于,所述图像为红外图像。
- 根据权利要求17-32任一项所述的图像处理装置,其特征在于,所述装置还包括红外传感器,所述待去噪图像通过所述红外传感器采集得到。
- 根据权利要求17-33任一项所述的图像处理装置,其特征在于,所述装置用于无人机。
- 一种计算机可读存储介质,其特征在于,所述计算机可读存储介质上存储有计算机程序,所述计算机程序被处理器执行时实现如权利要求1至16任一项所述图像处理方法。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2019/121987 WO2021102913A1 (zh) | 2019-11-29 | 2019-11-29 | 图像处理方法、装置及存储介质 |
CN201980049929.4A CN112513936A (zh) | 2019-11-29 | 2019-11-29 | 图像处理方法、装置及存储介质 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/CN2019/121987 WO2021102913A1 (zh) | 2019-11-29 | 2019-11-29 | 图像处理方法、装置及存储介质 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2021102913A1 true WO2021102913A1 (zh) | 2021-06-03 |
Family
ID=74923730
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2019/121987 WO2021102913A1 (zh) | 2019-11-29 | 2019-11-29 | 图像处理方法、装置及存储介质 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN112513936A (zh) |
WO (1) | WO2021102913A1 (zh) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113344820A (zh) * | 2021-06-28 | 2021-09-03 | Oppo广东移动通信有限公司 | 图像处理方法及装置、计算机可读介质、电子设备 |
CN114708248A (zh) * | 2022-04-22 | 2022-07-05 | 中广核风电有限公司 | 海底电缆状态监测数据压缩方法、装置及电子设备 |
CN116051409A (zh) * | 2023-01-09 | 2023-05-02 | 长春理工大学 | 一种非制冷红外探测器的最优偏置电压控制方法 |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113538265A (zh) * | 2021-07-06 | 2021-10-22 | Oppo广东移动通信有限公司 | 图像去噪方法及装置、计算机可读介质、电子设备 |
CN115619652A (zh) * | 2021-07-15 | 2023-01-17 | 浙江宇视科技有限公司 | 图像盲去噪方法、装置、电子设备和存储介质 |
CN116563175B (zh) * | 2023-07-12 | 2023-09-15 | 中国科学院长春光学精密机械与物理研究所 | 光电成像设备系统固定噪声去除方法、装置和存储介质 |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110311103A1 (en) * | 2009-02-17 | 2011-12-22 | Autoliv Developement Ab | method and system of automatically detecting objects in front of a motor vehicle |
CN102509269A (zh) * | 2011-11-10 | 2012-06-20 | 重庆工业职业技术学院 | 一种结合曲线波的基于图像子块相似性的图像去噪方法 |
CN104036471A (zh) * | 2013-03-05 | 2014-09-10 | 腾讯科技(深圳)有限公司 | 一种图像噪声估值方法及图像噪声估值装置 |
CN110390650A (zh) * | 2019-07-23 | 2019-10-29 | 中南大学 | 基于密集连接和生成对抗网络的oct图像去噪方法 |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107645621A (zh) * | 2016-07-20 | 2018-01-30 | 阿里巴巴集团控股有限公司 | 一种视频处理的方法和设备 |
-
2019
- 2019-11-29 WO PCT/CN2019/121987 patent/WO2021102913A1/zh active Application Filing
- 2019-11-29 CN CN201980049929.4A patent/CN112513936A/zh active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20110311103A1 (en) * | 2009-02-17 | 2011-12-22 | Autoliv Developement Ab | method and system of automatically detecting objects in front of a motor vehicle |
CN102509269A (zh) * | 2011-11-10 | 2012-06-20 | 重庆工业职业技术学院 | 一种结合曲线波的基于图像子块相似性的图像去噪方法 |
CN104036471A (zh) * | 2013-03-05 | 2014-09-10 | 腾讯科技(深圳)有限公司 | 一种图像噪声估值方法及图像噪声估值装置 |
CN110390650A (zh) * | 2019-07-23 | 2019-10-29 | 中南大学 | 基于密集连接和生成对抗网络的oct图像去噪方法 |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113344820A (zh) * | 2021-06-28 | 2021-09-03 | Oppo广东移动通信有限公司 | 图像处理方法及装置、计算机可读介质、电子设备 |
CN113344820B (zh) * | 2021-06-28 | 2024-05-10 | Oppo广东移动通信有限公司 | 图像处理方法及装置、计算机可读介质、电子设备 |
CN114708248A (zh) * | 2022-04-22 | 2022-07-05 | 中广核风电有限公司 | 海底电缆状态监测数据压缩方法、装置及电子设备 |
CN116051409A (zh) * | 2023-01-09 | 2023-05-02 | 长春理工大学 | 一种非制冷红外探测器的最优偏置电压控制方法 |
CN116051409B (zh) * | 2023-01-09 | 2024-06-07 | 长春理工大学 | 一种非制冷红外探测器的最优偏置电压控制方法 |
Also Published As
Publication number | Publication date |
---|---|
CN112513936A (zh) | 2021-03-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2021102913A1 (zh) | 图像处理方法、装置及存储介质 | |
CN110717942B (zh) | 图像处理方法和装置、电子设备、计算机可读存储介质 | |
US8582820B2 (en) | Coded aperture camera with adaptive image processing | |
US8290305B2 (en) | Registration of 3D point cloud data to 2D electro-optical image data | |
US20150117756A1 (en) | Processing of Light Fields by Transforming to Scale and Depth Space | |
CN113077459B (zh) | 图像清晰度检测方法、装置、电子设备及存储介质 | |
US20170308998A1 (en) | Motion Image Compensation Method and Device, Display Device | |
US20110267485A1 (en) | Range measurement using a coded aperture | |
US8948453B2 (en) | Device, method and non-transitory computer readable storage medium for detecting object | |
JP6830712B1 (ja) | ランダムサンプリング一貫性に基づく魚眼画像の有効領域抽出方法 | |
CN108205657A (zh) | 视频镜头分割的方法、存储介质和移动终端 | |
WO2023060927A1 (zh) | 一种3d光栅检测方法、装置、计算机设备及可读存储介质 | |
WO2021134642A1 (zh) | 图像处理方法、装置及存储介质 | |
CN111179182B (zh) | 图像的处理方法及装置、存储介质及处理器 | |
US11120572B2 (en) | Method, system and apparatus for associating a target object in images | |
CN113727095A (zh) | 摄像头移动的检测方法、装置、设备、系统及存储介质 | |
CN112950468A (zh) | 图像拼接方法、电子设备及可读存储介质 | |
US20180357784A1 (en) | Method for characterising a scene by computing 3d orientation | |
CN116958058A (zh) | 镜头脏污检测方法、装置及图像检测设备 | |
CN113409375B (zh) | 图像处理方法、装置以及非易失性存储介质 | |
KR102288237B1 (ko) | 이미지 연관 방법, 시스템 및 장치 | |
WO2022165672A1 (zh) | 点云处理方法、装置和计算机可读存储介质 | |
CN114820376A (zh) | 条带噪声的融合校正方法、装置、电子设备及存储介质 | |
CN113838012A (zh) | Mura检测方法及装置、计算机可读存储介质、终端 | |
WO2017080236A1 (zh) | 图像处理方法及装置 |
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: 19954657 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 19954657 Country of ref document: EP Kind code of ref document: A1 |