WO2020042303A1 - 一种识别图像差异的方法及装置 - Google Patents

一种识别图像差异的方法及装置 Download PDF

Info

Publication number
WO2020042303A1
WO2020042303A1 PCT/CN2018/111165 CN2018111165W WO2020042303A1 WO 2020042303 A1 WO2020042303 A1 WO 2020042303A1 CN 2018111165 W CN2018111165 W CN 2018111165W WO 2020042303 A1 WO2020042303 A1 WO 2020042303A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
difference
matching
pixel point
feature
Prior art date
Application number
PCT/CN2018/111165
Other languages
English (en)
French (fr)
Inventor
党静
王雅儒
许龙
Original Assignee
深圳开立生物医疗科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳开立生物医疗科技股份有限公司 filed Critical 深圳开立生物医疗科技股份有限公司
Publication of WO2020042303A1 publication Critical patent/WO2020042303A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00Arrangements for image or video recognition or understanding
    • G06V10/70Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/751Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06VIMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00Indexing scheme relating to image or video recognition or understanding
    • G06V2201/03Recognition of patterns in medical or anatomical images

Definitions

  • the present invention relates to the technical field of digital image processing, and in particular, to a method and device for identifying image differences.
  • Medical imaging refers to the technology and process of obtaining an internal tissue image of a human body or a part of the human body in a non-invasive manner for medical or medical research.
  • the application of medical imaging makes it easy for doctors to observe the internal image or structure of the patient's body or body part location.
  • Doctors usually compare and observe the medical images of a certain part of the patient's body at different periods to determine whether the body part of the patient has changed. Due to the complexity of human tissue and the corresponding imaging content, doctors need to make a careful comparison to accurately determine whether there are differences in the medical images of the same body part in different periods. In addition, patients have a long treatment cycle and a large number of imaging. At the same time, more and more patients rely on medical images to observe the physical condition. Doctors rely on the naked eye to observe whether there are differences in medical images of the same body part at different periods. Doctors are less efficient at observing medical image differences with the naked eye.
  • the present invention proposes a method and device for identifying image differences, which can realize automatic identification of medical image differences.
  • a method for identifying image differences including:
  • a difference image region of the first image and the second image is determined.
  • the acquiring the first image and the second image of the difference to be identified includes:
  • Feature pixels of each of the first image and the comparison image set are separately extracted to obtain the feature pixel points of the first image and the features of each image in the comparison image set Pixel set
  • the calculating the difference image between the first image and the second image according to the coordinates of matching pixel points of the first image and the second image includes:
  • determining a difference image area between the first image and the second image based on the difference image includes:
  • Connected-domain labeling processing is performed on the mask map to determine a difference image area between the first image and the second image.
  • the method further includes:
  • the area and diameter of the difference image area are calculated and output.
  • a device for identifying image differences includes:
  • An image acquisition unit configured to acquire a first image and a second image of a difference to be identified
  • a feature pixel point obtaining unit configured to obtain a feature pixel point set of the first image and the second image respectively;
  • a matching processing unit configured to perform matching processing on the characteristic pixel point set of the first image and the characteristic pixel point set of the second image, and determine coordinates of matching pixel points of the first image and the second image ;
  • a difference calculation unit configured to calculate a difference image of the first image and the second image according to coordinates of matching pixel points of the first image and the second image;
  • a difference image determining unit is configured to determine a difference image area of the first image and the second image according to the difference image.
  • the image acquisition unit includes:
  • a first obtaining unit configured to obtain a first image and a preset contrast image set; wherein the images in the contrast image set are the same as the imaging target of the first image;
  • Feature pixel point extraction unit configured to extract feature pixel points of each of the first image and the contrast image set separately, to obtain the feature pixel point set of the first image, and the contrast image set A set of characteristic pixels of each image in the image;
  • An image selection unit configured to match the feature pixel point set of the first image with the feature pixel point set of each image in the comparison image set, and select from the comparison image set the The image with the most matching feature pixels in the first image is used as the second image.
  • the difference calculation unit includes:
  • a first calculation unit configured to calculate a transformation matrix of the first image or the second image according to coordinates of matching pixel points of the first image and the second image;
  • An image conversion unit configured to convert the first image and the second image into images displayed at the same coordinate position according to the transformation matrix
  • a second calculation unit is configured to calculate a difference between the first image and the second image displayed at the same position, and perform pixel normalization processing to obtain a difference image between the first image and the second image.
  • the difference image determination unit includes:
  • An image inference unit configured to input the difference image into a trained convolutional neural network, and cause the convolutional neural network to infer to obtain a mask image of the difference image; wherein the mask image includes the mask image The effective difference content of the difference image;
  • a labeling processing unit is configured to perform connected-field labeling processing on the mask map to determine a difference image area between the first image and the second image.
  • the device further includes:
  • An output unit is configured to perform contour tracing processing on the difference image region to obtain the outline of the difference image region; and calculate and output the area and diameter of the difference image region.
  • the characteristic pixel points of the first image and the second image are respectively acquired; and then the characteristic pixel points of the first image are obtained.
  • Performing matching processing on the set and the feature pixel set of the second image determining coordinates of matching pixel points of the first image and the second image, and according to matching between the first image and the second image
  • the coordinates of the pixel points are calculated to obtain a difference image between the first image and the second image; and finally, a difference image region of the first image and the second image is determined according to the difference image.
  • the above technical solution realizes the automatic recognition of the difference image area between two images, that is, the automatic recognition of the difference between the images, which is applied to the medical image difference recognition, is faster than the doctor's naked eye to identify the medical image difference, and the recognition efficiency is more efficient. high.
  • FIG. 1 is a schematic flowchart of a method for identifying image differences according to an embodiment of the present invention
  • FIG. 2 is a schematic flowchart of another method for identifying image differences according to an embodiment of the present invention
  • FIG. 3 is a schematic flowchart of still another method for identifying image differences according to an embodiment of the present invention.
  • FIG. 4 is a schematic flowchart of still another method for identifying image differences according to an embodiment of the present invention.
  • FIG. 5 is a schematic structural diagram of an apparatus for identifying image differences according to an embodiment of the present invention.
  • FIG. 6 is a schematic structural diagram of another apparatus for identifying image differences according to an embodiment of the present invention.
  • FIG. 7 is a schematic structural diagram of another apparatus for identifying image differences according to an embodiment of the present invention.
  • FIG. 8 is a schematic structural diagram of another apparatus for identifying image differences according to an embodiment of the present invention.
  • An embodiment of the present invention discloses a method for identifying image differences. Referring to FIG. 1, the method includes:
  • the above-mentioned first image and second image respectively refer to medical images taken for a same body part at different times during a patient's consultation process.
  • the first image is a medical image taken by a patient for a certain body part at the current moment
  • the second image is a medical image taken by the patient for the same body part in the medical history. It can be understood that the two images are compared to identify the difference , You can determine whether the same body part of the patient has changed, such as whether a lesion or a change in tissue structure has occurred.
  • the technical solution of the embodiment of the present invention may be implemented in software.
  • the above-mentioned process of obtaining the first image and the second image of the difference to be identified may be to provide a user with an image library image selection Window, so that the user can select an image from the image library that needs to identify the difference, and at the same time set an image display window, and at the same time display the image that the user chooses to identify the difference.
  • the user can also select multiple images for differential recognition.
  • the doctor selects multiple historical medical images for the same body part from the historical visit image database of the same patient for the above-mentioned current time. Differential recognition is performed on medical images of the same body part, or only differential recognition of historical medical images may be performed.
  • the differences between the first image and the second image are identified as an example to introduce the differences in the identified images according to the embodiment of the present invention
  • the method in any image display mode or in different scenes of different domain recognition scenarios, can refer to the method for identifying image differences proposed in the embodiments of the present invention to identify image differences, which are all within the protection scope of the embodiments of the present invention.
  • a characteristic pixel point of an image refers to a pixel point capable of representing an image characteristic.
  • the gray levels and edge gradients in the neighborhood of each pixel of the first image and the second image are respectively calculated, and then the pixels that can represent image features are selected according to the gray levels and edge gradients in the neighborhood of the pixels, as features pixel.
  • All the characteristic pixel points of the first image and the second image respectively constitute the characteristic pixel point set of the first image and the characteristic pixel point set of the second image.
  • the characteristic pixel point set of the first image may represent the first image and the second image.
  • the set of feature pixels may represent the second image.
  • each feature pixel point in the feature pixel point set of the first image is matched with each feature pixel point in the feature pixel point set of the second image, and the feature pixel points of the first image and the second image are determined. Whether the similarity of the characteristic pixel points is greater than a set threshold. If the similarity between the characteristic pixel points of the first image and the characteristic pixel points of the second image is greater than the set threshold, the characteristic pixel points of the first image and the second image are considered Feature pixel matching.
  • the matching feature pixel points in the feature pixel point set of the first image and the feature pixel point set of the second image are determined, that is, the matching pixel points of the first image and the second image are determined.
  • the coordinates of the matching pixel points of the first image and the second image in the image coordinate system are determined.
  • the first image and the second image are converted to the same coordinate position based on the coordinates of the matching pixel points of the first image and the second image. Display, and then subtract the first image and the second image displayed at the same coordinate position to obtain a difference image between the first image and the second image.
  • the effective difference in the difference image that is, the effective difference between the first image and the second image is inferred from the difference image, and then will be valid
  • the difference is labeled in the connected domain, and the difference image area is obtained.
  • the technical solution for automatically identifying image differences in the embodiment of the present invention obtains characteristic pixel points of the first image and the second image after acquiring the first image and the second image of the difference to be identified. Set; then performing matching processing on the feature pixel point set of the first image and the feature pixel point set of the second image, determining coordinates of matching pixel points of the first image and the second image, and according to The coordinates of matching pixel points of the first image and the second image are calculated to obtain a difference image between the first image and the second image; and finally, the first image is determined according to the difference image. A difference image area from the second image.
  • the above technical solution realizes the automatic recognition of the difference image area between two images, that is, the automatic recognition of the difference between the images, which is applied to the medical image difference recognition, is faster than the doctor's naked eye to identify the medical image difference, and the recognition efficiency is more efficient. high.
  • acquiring the first image and the second image of the difference to be identified includes:
  • the above-mentioned first image refers to a medical image taken by a patient for a certain body part at the current moment.
  • the preset comparison image set refers to a collection of medical images taken by the patient for the same body part in the medical history.
  • the comparison image set may be a medical image stored in a medical image database.
  • a characteristic pixel point of an image refers to a pixel point capable of representing an image characteristic.
  • the gray levels and edge gradients in the neighborhood of each pixel of each of the first image and each image in the above-mentioned comparison image set are calculated, and then according to the range of the gray levels and edge gradients in the neighborhood of the pixel, The pixels that can represent the characteristics of the image are selected as the characteristic pixels.
  • the feature pixel points of each of the first image and the comparison image set are extracted separately to obtain the feature pixel point set of the first image and each of the comparison image sets.
  • the grayscale and edge gradient values of each feature pixel point in the feature pixel point set of the first image are compared with each feature in the feature pixel point set of each image in the comparison image set, respectively.
  • the grayscale and edge gradient values of the pixels are compared, and the characteristic pixel points whose difference between the grayscale and edge gradient values are less than a set threshold are determined as matching characteristic pixel points.
  • the number of matching pixels of each of the first image and each of the comparison image sets is determined, and an image with the largest number of matching pixels of the first image from the comparison image set is selected as Second image.
  • the above-mentioned second image is essentially the image most similar to the first image, that is, the history medicine that is most similar to the first image currently taken for a certain body part of the patient and is taken for the same section of the body image.
  • an image in any period in which the number of matching feature pixels of the first image exceeds a certain threshold is used as the second image.
  • An image in any of the above periods is an image similar to the first image, but not necessarily the most similar image.
  • the second image selected at this time is selected for any historical period of the same body part of the patient Medical image.
  • Steps S204 to S207 in this embodiment correspond to steps S102 to S105 in the method embodiment shown in FIG. 1, respectively.
  • the first image and the image are calculated and calculated according to coordinates of matching pixel points of the first image and the second image.
  • the difference image of the second image includes:
  • the foregoing transformation matrix refers to a transformation matrix that translates an image display position.
  • the display position coordinates of one image are fixed, and then a transformation matrix that translates the other image to the same display position is calculated based on the coordinates of matching pixel points with the other image .
  • fixing the display position of the first image in the image coordinate system and then calculating the transformation matrix that translates the second image to the first image display position according to the coordinates of the matching pixel points of the first image and the second image, to obtain The transformation matrix of the second image.
  • the display position of the image corresponding to the transformation matrix is shifted, so that the two images described above are displayed at the same coordinate position.
  • the display position of the second image is transformed according to the transformation matrix, that is, the second image is translated to the same coordinate position as the display position of the first image for display, so that The first image and the second image become images displayed at the same coordinate position.
  • a difference operation is performed on the first image and the second image displayed at the same coordinate position, a difference between the first image and the second image is calculated, and the calculated pixel difference is normalized to obtain Difference image of the first image and the second image.
  • Steps S301 to S303 and S307 in this embodiment correspond to steps S101 to S103 and S105 in the method embodiment shown in FIG. 1, respectively.
  • For specific content refer to the content of the method embodiment shown in FIG. To repeat.
  • the determining a difference image area of the first image and the second image based on the difference image includes:
  • the embodiment of the present invention trains a convolutional neural network, inputs a difference image to the convolutional neural network, causes the convolutional neural network to infer a mask map of the input difference image, and then according to a previously determined input difference image
  • the mask map is used to adjust the convolutional neural network, and the above process is repeatedly performed to train the convolutional neural network, so that the convolutional neural network has the ability to infer the mask map of the difference image.
  • the above-mentioned mask map refers to an image in which pixel points of an image are represented in binary. Effective pixels in the image are marked as 1 and unnecessary pixels are marked as 0.
  • the specific display content of the mask image of the difference image is as follows: the value of the pixel point in the difference image indicating that the first image and the second image are different is 1, which indicates that the first image and the second image have the same pixel point The value is 0. It can be understood that in the mask image of the difference image, a pixel value of 1 indicates that the corresponding pixel points of the first image and the second image are different, and a pixel value of 0 indicates a correspondence between the first image and the second image. The pixels are the same.
  • the difference image of the first image and the second image is input to the trained convolutional neural network, and a mask map of the difference image is obtained by reasoning.
  • the obtained connected domain represents the first image and the second image.
  • the area of the difference portion of the image that is, the difference image area of the first image and the second image.
  • Steps S401 to S404 in this embodiment correspond to steps S101 to S104 in the method embodiment shown in FIG. 1, respectively.
  • the foregoing method for identifying image differences further includes:
  • the area and diameter of the difference image area are calculated and output.
  • the present invention after determining the difference image area of the first image and the second image, further performing contour tracing processing on the determined difference image area, that is, drawing the outline of the difference image area.
  • the outline of the difference image area is directly drawn on the first image or the second image displayed at the same coordinate position.
  • the embodiment of the present invention may also calculate the area and diameter of the image area included in the drawn difference image contour. Specifically, by calculating the number of pixels marked as 1 in each direction included in the difference image area, the diameter of the contour of the difference image area in that direction can be determined; the number of all pixels marked as 1 included in the difference image area is counted, The area of the image area included in the outline of the difference image area can be determined.
  • contour tracing processing on the difference image region obtains the outline of the difference image region, and the specific processing procedure for calculating the area and diameter of the difference image region. See also the commonly used processing procedure for calculating the parameters of the image region of interest .
  • the embodiment of the present invention can further control and output the contour, area, and diameter, so that the user (doctor) can more clearly discover the patient's body Changes.
  • the device includes:
  • An image acquisition unit 100 configured to acquire a first image and a second image of a difference to be identified
  • a feature pixel point obtaining unit 110 configured to obtain a feature pixel point set of the first image and the second image respectively;
  • the matching processing unit 120 is configured to perform matching processing on the characteristic pixel point set of the first image and the characteristic pixel point set of the second image, and determine the matching pixel points of the first image and the second image. coordinate;
  • a difference calculation unit 130 configured to calculate and obtain a difference image of the first image and the second image according to coordinates of matching pixel points of the first image and the second image;
  • a difference image determination unit 140 is configured to determine a difference image region of the first image and the second image according to the difference image.
  • the image acquisition unit 100 includes:
  • a first obtaining unit 1001 configured to obtain a first image and a preset contrast image set; wherein the images in the contrast image set are the same as the imaging target of the first image;
  • a feature pixel extraction unit 1002 is configured to extract feature pixels of each of the first image and each of the comparison image sets, to obtain a feature pixel set of the first image, and the comparison image.
  • An image selection unit 1003 is configured to match the feature pixel point set of the first image with the feature pixel point set of each image in the comparison image set, and select the feature pixel point set from the comparison image set. The image with the most matching feature pixels in the first image is used as the second image.
  • the difference calculation unit 130 includes:
  • a first calculation unit 1301, configured to calculate a transformation matrix of the first image or the second image according to coordinates of matching pixel points of the first image and the second image;
  • An image conversion unit 1302, configured to convert the first image and the second image into images displayed at the same coordinate position according to the transformation matrix
  • a second calculation unit 1303 is configured to calculate a difference between the first image and the second image displayed at the same position, and perform pixel normalization processing to obtain a difference image between the first image and the second image.
  • the difference image determining unit 140 includes:
  • An image inference unit 1401 is configured to input the difference image into a trained convolutional neural network, and cause the convolutional neural network to infer to obtain a mask image of the difference image; wherein the mask image includes the mask image of the difference image. Describe the effective difference content of the difference image;
  • a labeling processing unit 1402 is configured to perform connected-field labeling processing on the mask map to determine a difference image area between the first image and the second image.
  • the device further includes:
  • An output unit is configured to perform contour tracing processing on the difference image region to obtain the outline of the difference image region; and calculate and output the area and diameter of the difference image region.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Computing Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)
  • Apparatus For Radiation Diagnosis (AREA)
  • Image Processing (AREA)

Abstract

本发明提出一种识别图像差异的方法,包括:获取待识别差异的第一图像和第二图像;分别获取所述第一图像和所述第二图像的特征像素点集合;对所述第一图像的特征像素点集合和所述第二图像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配像素点的坐标;根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像;根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。上述技术方案能够实现自动识别图像差异,相对有人工识别图像差异,识别效率更高。

Description

一种识别图像差异的方法及装置
本申请要求于2018年8月27日提交中国专利局、申请号为201810983018.6、发明名称为“一种识别图像差异的方法及装置”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本发明涉及数字图像处理技术领域,尤其涉及一种识别图像差异的方法及装置。
背景技术
医学影像是指为了医疗或医学研究,对人体或人体某部分,以非侵入方式取得内部组织影像的技术与处理过程。医学影像的应用方便了医生观察患者身体或身体部分位置的内部图像或结构。
医生通常通过将患者身体某一部位的不同时期的医学图像进行对比观察,判断患者的该身体部分是否发生变化。由于人体组织复杂,成像内容也相应复杂,对于医生来说需要很仔细地对比才能准确判断不同时期同一身体部位的医学图像是否存在差异。加之患者治疗周期长,成像数量多,同时越来越多的患者依靠医学影像观察身体状况,医生依靠肉眼观察不同时期同一身体部位的医学图像是否存在差异是一项繁重的劳动付出,并且完全依靠医生肉眼观察医学图像差异的效率较低。
发明内容
基于上述现有技术的缺陷和不足,本发明提出一种识别图像差异的方法及装置,能够实现自动化地识别医学图像差异。
一种识别图像差异的方法,包括:
获取待识别差异的第一图像和第二图像;
分别获取所述第一图像和所述第二图像的特征像素点集合;
对所述第一图像的特征像素点集合和所述第二图像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配像素点的坐标;
根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第 一图像和所述第二图像的差值图像;
根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。
可选的,所述获取待识别差异的第一图像和第二图像,包括:
获取第一图像和预设的对比图像集合;其中,所述对比图像集合中的图像与所述第一图像的成像目标相同;
分别提取所述第一图像和所述对比图像集合中的每一幅图像的特征像素点,得到所述第一图像的特征像素点集合,以及所述对比图像集合中的每一幅图像的特征像素点集合;
分别将所述第一图像的特征像素点集合与所述对比图像集合中的每一幅图像的特征像素点集合进行匹配,从所述对比图像集合中选取出与所述第一图像的匹配特征像素点最多的图像,作为第二图像。
可选的,所述根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像,包括:
根据所述第一图像和所述第二图像的匹配像素点的坐标,计算所述第一图像或所述第二图像的变换矩阵;
根据所述变换矩阵,将所述第一图像和所述第二图像转换为在相同坐标位置显示的图像;
计算在相同位置显示的第一图像和第二图像的差值,并进行像素归一化处理,得到所述第一图像和所述第二图像的差值图像。
可选的,所述根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域,包括:
将所述差值图像输入经过训练的卷积神经网络,使所述卷积神经网络推理得到所述差值图像的掩码图;其中,所述掩码图包含所述差值图像的有效差异内容;
对所述掩码图进行连通域标记处理,确定所述第一图像和所述第二图像的差异图像区域。
可选的,所述方法还包括:
对所述差异图像区域进行轮廓描迹处理,得到所述差异图像区域的轮廓;
计算并输出所述差异图像区域的面积和直径。
一种识别图像差异的装置,包括:
图像获取单元,用于获取待识别差异的第一图像和第二图像;
特征像素点获取单元,用于分别获取所述第一图像和所述第二图像的特征像素点集合;
匹配处理单元,用于对所述第一图像的特征像素点集合和所述第二图像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配像素点的坐标;
差值计算单元,用于根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像;
差异图像确定单元,用于根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。
可选的,所述图像获取单元,包括:
第一获取单元,用于获取第一图像和预设的对比图像集合;其中,所述对比图像集合中的图像与所述第一图像的成像目标相同;
特征像素点提取单元,用于分别提取所述第一图像和所述对比图像集合中的每一幅图像的特征像素点,得到所述第一图像的特征像素点集合,以及所述对比图像集合中的每一幅图像的特征像素点集合;
图像选择单元,用于分别将所述第一图像的特征像素点集合与所述对比图像集合中的每一幅图像的特征像素点集合进行匹配,从所述对比图像集合中选取出与所述第一图像的匹配特征像素点最多的图像,作为第二图像。
可选的,所述差值计算单元,包括:
第一计算单元,用于根据所述第一图像和所述第二图像的匹配像素点的坐标,计算所述第一图像或所述第二图像的变换矩阵;
图像转换单元,用于根据所述变换矩阵,将所述第一图像和所述第二图像转换为在相同坐标位置显示的图像;
第二计算单元,用于计算在相同位置显示的第一图像和第二图像的差值,并进行像素归一化处理,得到所述第一图像和所述第二图像的差值图像。
可选的,所述差异图像确定单元,包括:
图像推理单元,用于将所述差值图像输入经过训练的卷积神经网络,使所述卷积神经网络推理得到所述差值图像的掩码图;其中,所述掩码图包含所述差值图像的有效差异内容;
标记处理单元,用于对所述掩码图进行连通域标记处理,确定所述第一图像和所述第二图像的差异图像区域。
可选的,所述装置还包括:
输出单元,用于对所述差异图像区域进行轮廓描迹处理,得到所述差异图像区域的轮廓;计算并输出所述差异图像区域的面积和直径。
本发明技术方案,在获取待识别差异的第一图像和第二图像后,分别获取所述第一图像和所述第二图像的特征像素点集合;然后对所述第一图像的特征像素点集合和所述第二图像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配像素点的坐标,以及根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像;最后根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。上述技术方案实现了自动化的识别两幅图像的差异图像区域,即实现了自动化的识别图像差异,将其应用到医学图像差异识别中,比医生肉眼识别医学图像差异的速度更快,识别效率更高。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案,下面将对实施例或现有技术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发明的实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据提供的附图获得其他的附图。
图1是本发明实施例提供的一种识别图像差异的方法的流程示意图;
图2是本发明实施例提供的另一种识别图像差异的方法的流程示意图;
图3是本发明实施例提供的的又一种识别图像差异的方法的流程示意图;
图4是本发明实施例提供的的再一种识别图像差异的方法的流程示意图;
图5是本发明实施例提供的一种识别图像差异的装置的结构示意图;
图6是本发明实施例提供的另一种识别图像差异的装置的结构示意图;
图7是本发明实施例提供的又一种识别图像差异的装置的结构示意图;
图8是本发明实施例提供的再一种识别图像差异的装置的结构示意图。
具体实施方式
下面将结合本发明实施例中的附图,对本发明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发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。
本发明实施例公开了一种识别图像差异的方法,参见图1所示,该方法包括:
S101、获取待识别差异的第一图像和第二图像;
具体的,上述的第一图像和第二图像,分别是指在患者就诊过程中不同时期针对同一身体部位拍摄的医学图像。例如上述第一图像为患者在当前时刻针对某一身体部位拍摄的医学图像,上述第二图像为患者在就诊历史中针对同一身体部位拍摄的医学图像,可以理解,将两幅图像进行对比识别差异,就可以判断该患者的上述同一身体部位是否发生变化,例如是否发生病变或组织结构发生变化等。
本发明实施例技术方案可以以软件形式实现,在实现本发明实施例技术方案的软件运行时,上述获取待识别差异的第一图像和第二图像的过程,可以是为用户提供图像库图像选择窗口,使用户从图像库中选择需要识别差异的图像,同时设置图像显示窗口,同时显示用户选择的需要识别差异的图像。更进一步的,用户还可以选择多幅图像进行差异识别,例如医生从同一患者的历史就诊图像资料库中,选择针对同一身体部位的多幅历史医学图像,用于与选择的当前时刻的针对上述同一身体部位的医学图像进行差异识别,或者也可以只进行对历史医学图像的差异识别。
可以理解,识别多幅图像差异时,也是每两幅图像之间对比识别差异,因此本发明实施例以识别上述第一图像和第二图像的差异为例介绍本发明实施例提出的识别图像差异的方法,在任意图像显示模式或不同领域图像的差异识别场景下,都可以参照本发明实施例提出的识别图像差异的方法识别图像差异,都在本发明实施例保护范围内。
S102、分别获取所述第一图像和所述第二图像的特征像素点集合;
具体的,图像的特征像素点,是指能够表示图像特征的像素点。本发明实施例分别计算第一图像和第二图像的各个像素点邻域内的灰阶、边缘梯度,然后根据像素点邻域内的灰阶、边缘梯度选取其中可以表示图像特征的像素点, 作为特征像素点。第一图像和第二图像的所有特征像素点分别构成第一图像的特征像素点集合和第二图像的特征像素点集合,则第一图像的特征像素点集合可以代表第一图像,第二图像的特征像素点集合可以代表第二图像。
S103、对所述第一图像的特征像素点集合和所述第二图像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配像素点的坐标;
具体的,将第一图像的特征像素点集合中的各个特征像素点分别与第二图像的特征像素点集合中的各个特征像素点进行匹配,判断第一图像的特征像素点与第二图像的特征像素点的相似度是否大于设定阈值,如果第一图像的特征像素点与第二图像的特征像素点的相似度大于设定阈值,则认为第一图像的特征像素点与第二图像的特征像素点匹配。
按照上述方法,分别确定第一图像的特征像素点集合与第二图像的特征像素点集合中的,相匹配的特征像素点,即确定第一图像与第二图像的匹配像素点。
然后,确定上述第一图像和第二图像的匹配像素点在图像坐标系中的坐标。
S104、根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像;
具体的,在确定第一图像和第二图像的匹配像素点的坐标后,以第一图像和第二图像的匹配像素点的坐标为基准,将第一图像和第二图像转换到同一坐标位置显示,然后将同一坐标位置显示的第一图像和第二图像求差,得到第一图像和第二图像的差值图像。
S105、根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。
具体的,在计算得到第一图像和第二图像的差值图像后,从差值图像中推理得到差值图像中的有效差异,也就是第一图像和第二图像的有效差异,然后将有效差异进行连通域标记,即得到差异图像区域。
通过上述介绍可见,本发明实施例自动化的识别图像差异的技术方案,在获取待识别差异的第一图像和第二图像后,分别获取所述第一图像和所述第二图像的特征像素点集合;然后对所述第一图像的特征像素点集合和所述第二图像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配 像素点的坐标,以及根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像;最后根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。上述技术方案实现了自动化的识别两幅图像的差异图像区域,即实现了自动化的识别图像差异,将其应用到医学图像差异识别中,比医生肉眼识别医学图像差异的速度更快,识别效率更高。
可选的,在本发明的另一个实施例中,参见图2所示,所述获取待识别差异的第一图像和第二图像,包括:
S201、获取第一图像和预设的对比图像集合;其中,所述对比图像集合中的图像与所述第一图像的成像目标相同;
具体的,上述第一图像,是指患者在当前时刻针对某一身体部位拍摄的医学图像。上述预设的对比图像集合,是指上述患者在就诊历史中,针对同一身体部位拍摄的医学图像的集合。该对比图像集合,可以是存储在医学图像资料库中的医学图像。
S202、分别提取所述第一图像和所述对比图像集合中的每一幅图像的特征像素点,得到所述第一图像的特征像素点集合,以及所述对比图像集合中的每一幅图像的特征像素点集合;
具体的,图像的特征像素点,是指能够表示图像特征的像素点。本发明实施例分别计算第一图像和上述对比图像集合中的每一幅图像的各个像素点邻域内的灰阶、边缘梯度,然后根据像素点邻域内的灰阶、边缘梯度的取值范围,选取其中可以表示图像特征的像素点,作为特征像素点。
按照上述的特征像素点选取方法,分别提取上述第一图像和上述对比图像集合中的每一幅图像的特征像素点,得到第一图像的特征像素点集合,和对比图像集合中的每一幅图像的特征像素点集合。
S203、分别将所述第一图像的特征像素点集合与所述对比图像集合中的每一幅图像的特征像素点集合进行匹配,从所述对比图像集合中选取出与所述第一图像的匹配特征像素点最多的图像,作为第二图像。
具体的,将上述第一图像的特征像素点集合中的每一个特征像素点的灰阶和边缘梯度值,分别与上述对比图像集合中的每一幅图像的特征像素点集合中的每一个特征像素点的灰阶和边缘梯度值进行对比,将灰阶和边缘梯度值的差 值小于设定阈值的特征像素点确定为匹配特征像素点。
按照上述方法,分别确定上述第一图像与上述对比图像集合中的每一幅图像的匹配像素点数量,从上述对比图像集合中选择出与上述第一图像的匹配像素点数量最多的图像,作为第二图像。
可以理解,上述第二图像实质上是与第一图像最相似的图像,也就是与当前针对患者某一身体部位拍摄的第一图像最相似的,针对同一身体部位拍摄的同一切面的历史医学图像。
进一步的,在分别提取到上述第一图像的特征像素点集合与上述对比图像集合中的每一幅图像的特征像素点集合后,可以参照上述的处理方法,从上述对比图像集合中,选择与第一图像的匹配特征像素点数量超过一定阈值的任意时期的一幅图像,作为第二图像。上述任意时期的一幅图像,则是与第一图像相似的图像,但并不一定是最相似的图像,此时选出的第二图像,是选取的针对患者同一身体部位的任意历史时期的医学图像。
本实施例中的步骤S204~S207分别对应图1所示的方法实施例中的步骤S102~S105,其具体内容请参见图1所示的方法实施例的内容,此处不再赘述。
可选的,在本发明的另一个实施例中,参见图3所示,所述根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像,包括:
S304、根据所述第一图像和所述第二图像的匹配像素点的坐标,计算所述第一图像或所述第二图像的变换矩阵;
具体的,上述变换矩阵,是指对图像显示位置进行平移的变换矩阵。对于上述的第一图像和第二图像,将其中一幅图像的显示位置坐标固定,然后根据与另一幅图像的匹配像素点的坐标,计算将另一幅图像平移到相同显示位置的变换矩阵。例如,将第一图像在图像坐标系的显示位置固定,然后根据第一图像和第二图像的匹配像素点坐标,计算将第二图像平移到第一图像显示位置进行显示的变换矩阵,即得到第二图像的变换矩阵。
S305、根据所述变换矩阵,将所述第一图像和所述第二图像转换为在相同坐标位置显示的图像;
具体的,根据上述计算得到的变换矩阵,将该变换矩阵对应的图像进行显示位置平移,使上述的两幅图像在相同的坐标位置显示。例如,通过步骤S304 计算得到第二图像的变换矩阵后,将第二图像按照该变换矩阵进行显示位置变换,也就是将第二图像平移到第一图像的显示位置相同的坐标位置进行显示,使第一图像和第二图像成为在相同坐标位置显示的图像。
S306、计算在相同位置显示的第一图像和第二图像的差值,并进行像素归一化处理,得到所述第一图像和所述第二图像的差值图像。
具体的,将上述在相同坐标位置显示的第一图像和第二图像进行差值运算,计算第一图像和第二图像的差值,并对计算得到的像素差值进行归一化处理,得到第一图像和第二图像的差值图像。
本实施例中的步骤S301~S303、S307分别对应图1所示的方法实施例中的步骤S101~S103、S105,其具体内容请参见图1所示的方法实施例的内容,此处不再赘述。
可选的,在本发明的另一个实施例中,参见图4所示,所述根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域,包括:
S405、将所述差值图像输入经过训练的卷积神经网络,使所述卷积神经网络推理得到所述差值图像的掩码图;其中,所述掩码图包含所述差值图像的有效差异内容;
具体的,本发明实施例训练卷积神经网络,对卷积神经网络输入差值图像,使卷积神经网络推理输入的差值图像的掩码图,然后根据事先已确定的输入的差值图像的掩码图对卷积神经网络进行调校,反复执行上述过程实现对卷积神经网络的训练,使该卷积神经网络具备推理差值图像的掩码图的能力。
上述的掩码图,是指以二进制表示图像像素点的图像,图像中的有效像素被标记为1,无用像素被标记为0。则上述的差值图像的掩码图的具体显示内容是:差值图像中表示第一图像和第二图像存在差异的像素点的值为1,表示第一图像和第二图像相同的像素点的值为0。可以理解,在上述差值图像的掩码图中,值为1的像素点表示第一图像和第二图像的对应像素点不同,值为0的像素点表示第一图像和第二图像的对应像素点相同。
本发明实施例将第一图像和第二图像的差值图像输入上述经过训练的卷积神经网络,推理得到该差值图像的掩码图。
S406、对所述掩码图进行连通域标记处理,确定所述第一图像和所述第二图像的差异图像区域。
具体的,将掩码图中,被标记为1的所有像素点进行连通域标记,实现将连通的被标记为1的像素点形成区域,则得到的连通域即为表示第一图像和第二图像的差异部分的区域,即第一图像和第二图像的差异图像区域。
本实施例中的步骤S401~S404分别对应图1所示的方法实施例中的步骤S101~S104,其具体内容请参见图1所示的方法实施例的内容,此处不再赘述。
可选的,在本发明的另一个实施例中还公开了,上述的识别图像差异的方法还包括:
对所述差异图像区域进行轮廓描迹处理,得到所述差异图像区域的轮廓;
计算并输出所述差异图像区域的面积和直径。
具体的,在本发明实施例中,在确定第一图像和第二图像的差异图像区域后,还进一步对确定的差异图像区域进行轮廓描迹处理,即描绘出差异图像区域的轮廓,具体可在相同坐标位置显示的第一图像或第二图像上直接描绘差异图像区域的轮廓。
进一步的,本发明实施例还可以计算所描绘出的差异图像轮廓所包含的图像区域的面积和直径。具体的,通过计算差异图像区域包含的各个方向的被标记为1的像素数量,即可确定该差异图像区域轮廓在该方向的直径;统计差异图像区域包含的所有被标记为1的像素数量,可以确定该差异图像区域轮廓所包含的图像区域的面积。
需要说明的是,上述对差异图像区域进行轮廓描迹处理得到差异图像区域的轮廓,以及计算差异图像区域面积和直径的具体处理过程,还可以参见常用的计算图像感兴趣区域的参数的处理过程。
在计算得到上述的差异图像区域的轮廓、差异图像区域的面积和直径后,本发明实施例还可以进一步控制输出上述的轮廓、面积和直径,以使用户(医生)可以更明确地发现患者身体的变化情况。
本发明另一实施例还公开了一种识别图像差异的装置,参见图5所示,该装置包括:
图像获取单元100,用于获取待识别差异的第一图像和第二图像;
特征像素点获取单元110,用于分别获取所述第一图像和所述第二图像的特征像素点集合;
匹配处理单元120,用于对所述第一图像的特征像素点集合和所述第二图 像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配像素点的坐标;
差值计算单元130,用于根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像;
差异图像确定单元140,用于根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。
可选的,在本发明的另一实施例中,参见图6所示,所述图像获取单元100,包括:
第一获取单元1001,用于获取第一图像和预设的对比图像集合;其中,所述对比图像集合中的图像与所述第一图像的成像目标相同;
特征像素点提取单元1002,用于分别提取所述第一图像和所述对比图像集合中的每一幅图像的特征像素点,得到所述第一图像的特征像素点集合,以及所述对比图像集合中的每一幅图像的特征像素点集合;
图像选择单元1003,用于分别将所述第一图像的特征像素点集合与所述对比图像集合中的每一幅图像的特征像素点集合进行匹配,从所述对比图像集合中选取出与所述第一图像的匹配特征像素点最多的图像,作为第二图像。
可选的,在本发明的另一实施例中,参见图7所示,所述差值计算单元130,包括:
第一计算单元1301,用于根据所述第一图像和所述第二图像的匹配像素点的坐标,计算所述第一图像或所述第二图像的变换矩阵;
图像转换单元1302,用于根据所述变换矩阵,将所述第一图像和所述第二图像转换为在相同坐标位置显示的图像;
第二计算单元1303,用于计算在相同位置显示的第一图像和第二图像的差值,并进行像素归一化处理,得到所述第一图像和所述第二图像的差值图像。
可选的,在本发明的另一实施例中,参见图8所示,所述差异图像确定单元140,包括:
图像推理单元1401,用于将所述差值图像输入经过训练的卷积神经网络,使所述卷积神经网络推理得到所述差值图像的掩码图;其中,所述掩码图包含所述差值图像的有效差异内容;
标记处理单元1402,用于对所述掩码图进行连通域标记处理,确定所述 第一图像和所述第二图像的差异图像区域。
可选的,在本发明的另一实施例中,所述装置还包括:
输出单元,用于对所述差异图像区域进行轮廓描迹处理,得到所述差异图像区域的轮廓;计算并输出所述差异图像区域的面积和直径。
具体的,上述各实施例中的各个单元的具体工作内容,请参见上述方法实施例的内容,此处不再赘述。
需要说明的是,本说明书中的各个实施例均采用递进的方式描述,每个实施例重点说明的都是与其他实施例的不同之处,各个实施例之间相同相似的部分互相参见即可。对于装置类实施例而言,由于其与方法实施例基本相似,所以描述的比较简单,相关之处参见方法实施例的部分说明即可。
最后,还需要说明的是,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。而且,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。
对所公开的实施例的上述说明,使本领域专业技术人员能够实现或使用本发明。对这些实施例的多种修改对本领域的专业技术人员来说将是显而易见的,本文中所定义的一般原理可以在不脱离本发明的精神或范围的情况下,在其它实施例中实现。因此,本发明将不会被限制于本文所示的这些实施例,而是要符合与本文所公开的原理和新颖特点相一致的最宽的范围。

Claims (10)

  1. 一种识别图像差异的方法,其特征在于,包括:
    获取待识别差异的第一图像和第二图像;
    分别获取所述第一图像和所述第二图像的特征像素点集合;
    对所述第一图像的特征像素点集合和所述第二图像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配像素点的坐标;
    根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像;
    根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。
  2. 根据权利要求1所述的方法,其特征在于,所述获取待识别差异的第一图像和第二图像,包括:
    获取第一图像和预设的对比图像集合;其中,所述对比图像集合中的图像与所述第一图像的成像目标相同;
    分别提取所述第一图像和所述对比图像集合中的每一幅图像的特征像素点,得到所述第一图像的特征像素点集合,以及所述对比图像集合中的每一幅图像的特征像素点集合;
    分别将所述第一图像的特征像素点集合与所述对比图像集合中的每一幅图像的特征像素点集合进行匹配,从所述对比图像集合中选取出与所述第一图像的匹配特征像素点最多的图像,作为第二图像。
  3. 根据权利要求1所述的方法,其特征在于,所述根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像,包括:
    根据所述第一图像和所述第二图像的匹配像素点的坐标,计算所述第一图像或所述第二图像的变换矩阵;
    根据所述变换矩阵,将所述第一图像和所述第二图像转换为在相同坐标位置显示的图像;
    计算在相同位置显示的第一图像和第二图像的差值,并进行像素归一化处理,得到所述第一图像和所述第二图像的差值图像。
  4. 根据权利要求1-3中任一权利要求所述的方法,其特征在于,所述根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域,包括:
    将所述差值图像输入经过训练的卷积神经网络,使所述卷积神经网络推理得到所述差值图像的掩码图;其中,所述掩码图包含所述差值图像的有效差异内容;
    对所述掩码图进行连通域标记处理,确定所述第一图像和所述第二图像的差异图像区域。
  5. 根据权利要求1-3中任一权利要求所述的方法,其特征在于,所述方法还包括:
    对所述差异图像区域进行轮廓描迹处理,得到所述差异图像区域的轮廓;
    计算并输出所述差异图像区域的面积和直径。
  6. 一种识别图像差异的装置,其特征在于,包括:
    图像获取单元,用于获取待识别差异的第一图像和第二图像;
    特征像素点获取单元,用于分别获取所述第一图像和所述第二图像的特征像素点集合;
    匹配处理单元,用于对所述第一图像的特征像素点集合和所述第二图像的特征像素点集合进行匹配处理,确定所述第一图像和所述第二图像的匹配像素点的坐标;
    差值计算单元,用于根据所述第一图像和所述第二图像的匹配像素点的坐标,计算得到所述第一图像和所述第二图像的差值图像;
    差异图像确定单元,用于根据所述差值图像,确定所述第一图像和所述第二图像的差异图像区域。
  7. 根据权利要求6所述的装置,其特征在于,所述图像获取单元,包括:
    第一获取单元,用于获取第一图像和预设的对比图像集合;其中,所述对比图像集合中的图像与所述第一图像的成像目标相同;
    特征像素点提取单元,用于分别提取所述第一图像和所述对比图像集合中的每一幅图像的特征像素点,得到所述第一图像的特征像素点集合,以及所述对比图像集合中的每一幅图像的特征像素点集合;
    图像选择单元,用于分别将所述第一图像的特征像素点集合与所述对比图像集合中的每一幅图像的特征像素点集合进行匹配,从所述对比图像集合中选取出与所述第一图像的匹配特征像素点最多的图像,作为第二图像。
  8. 根据权利要求6所述的装置,其特征在于,所述差值计算单元,包括:
    第一计算单元,用于根据所述第一图像和所述第二图像的匹配像素点的坐标,计算所述第一图像或所述第二图像的变换矩阵;
    图像转换单元,用于根据所述变换矩阵,将所述第一图像和所述第二图像转换为在相同坐标位置显示的图像;
    第二计算单元,用于计算在相同位置显示的第一图像和第二图像的差值,并进行像素归一化处理,得到所述第一图像和所述第二图像的差值图像。
  9. 根据权利要求6-8中任一权利要求所述的装置,其特征在于,所述差异图像确定单元,包括:
    图像推理单元,用于将所述差值图像输入经过训练的卷积神经网络,使所述卷积神经网络推理得到所述差值图像的掩码图;其中,所述掩码图包含所述差值图像的有效差异内容;
    标记处理单元,用于对所述掩码图进行连通域标记处理,确定所述第一图像和所述第二图像的差异图像区域。
  10. 根据权利要求6-8中任一权利要求所述的装置,其特征在于,所述装置还包括:
    输出单元,用于对所述差异图像区域进行轮廓描迹处理,得到所述差异图像区域的轮廓;计算并输出所述差异图像区域的面积和直径。
PCT/CN2018/111165 2018-08-27 2018-10-22 一种识别图像差异的方法及装置 WO2020042303A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201810983018.6 2018-08-27
CN201810983018.6A CN109242011A (zh) 2018-08-27 2018-08-27 一种识别图像差异的方法及装置

Publications (1)

Publication Number Publication Date
WO2020042303A1 true WO2020042303A1 (zh) 2020-03-05

Family

ID=65069290

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/111165 WO2020042303A1 (zh) 2018-08-27 2018-10-22 一种识别图像差异的方法及装置

Country Status (2)

Country Link
CN (1) CN109242011A (zh)
WO (1) WO2020042303A1 (zh)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109949898B (zh) * 2019-02-19 2024-01-05 东软医疗系统股份有限公司 文件上传方法、存储方法、下载方法、系统、装置及设备
CN110222590B (zh) * 2019-05-15 2021-05-25 北京字节跳动网络技术有限公司 图像差异判断方法、装置及电子设备
CN110458836A (zh) * 2019-08-16 2019-11-15 深圳开立生物医疗科技股份有限公司 一种超声造影成像方法、装置和设备及可读存储介质
CN111462190B (zh) * 2020-04-20 2023-11-17 海信集团有限公司 一种智能冰箱及食材录入方法
CN111913873A (zh) * 2020-06-17 2020-11-10 浙江数链科技有限公司 图片校验方法、装置、系统和计算机可读存储介质
CN112101272B (zh) * 2020-09-23 2024-05-14 阿波罗智联(北京)科技有限公司 交通灯检测的方法、装置、计算机存储介质和路侧设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102800074A (zh) * 2012-07-12 2012-11-28 西安电子科技大学 基于轮廓波变换的sar图像变化检测差异图生成方法
CN102930519A (zh) * 2012-09-18 2013-02-13 西安电子科技大学 基于非局部均值的sar图像变化检测差异图生成方法
CN103186897A (zh) * 2011-12-29 2013-07-03 北京大学 获取图像差异度量结果的方法及装置
CN104602025A (zh) * 2015-01-09 2015-05-06 华为技术有限公司 处理图像的方法和装置
CN104751444A (zh) * 2014-12-24 2015-07-01 张�林 一种从相似图像中筛选显著差异的方法及装置

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101398907B (zh) * 2007-09-26 2011-08-10 中国科学院自动化研究所 一种用于移动机器人的二维码结构及解码方法
TWI534687B (zh) * 2014-01-24 2016-05-21 原相科技股份有限公司 光學觸控系統及其物件分析方法
CN105976389A (zh) * 2016-05-20 2016-09-28 南京理工大学 一种手机底板连接器缺陷检测方法
CN106257497B (zh) * 2016-07-27 2020-05-08 中测高科(北京)测绘工程技术有限责任公司 一种图像同名点的匹配方法及装置
CN106355607B (zh) * 2016-08-12 2019-01-22 辽宁工程技术大学 一种宽基线彩色图像模板匹配方法
CN106846293B (zh) * 2016-12-14 2020-08-07 海纳医信(北京)软件科技有限责任公司 图像处理方法和装置
CN107679502B (zh) * 2017-10-12 2019-10-08 南京行者易智能交通科技有限公司 一种基于深度学习图像语义分割的人数估计方法
CN107895365B (zh) * 2017-11-01 2020-08-18 国网山东省电力公司电力科学研究院 一种输电通道外破防护的图像匹配方法及监测系统
CN108389207A (zh) * 2018-04-28 2018-08-10 上海视可电子科技有限公司 一种牙病诊断方法、诊断设备及智能图像采集装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103186897A (zh) * 2011-12-29 2013-07-03 北京大学 获取图像差异度量结果的方法及装置
CN102800074A (zh) * 2012-07-12 2012-11-28 西安电子科技大学 基于轮廓波变换的sar图像变化检测差异图生成方法
CN102930519A (zh) * 2012-09-18 2013-02-13 西安电子科技大学 基于非局部均值的sar图像变化检测差异图生成方法
CN104751444A (zh) * 2014-12-24 2015-07-01 张�林 一种从相似图像中筛选显著差异的方法及装置
CN104602025A (zh) * 2015-01-09 2015-05-06 华为技术有限公司 处理图像的方法和装置

Also Published As

Publication number Publication date
CN109242011A (zh) 2019-01-18

Similar Documents

Publication Publication Date Title
WO2020042303A1 (zh) 一种识别图像差异的方法及装置
Rajinikanth et al. Otsu's multi-thresholding and active contour snake model to segment dermoscopy images
US20200303062A1 (en) Medical image aided diagnosis method and system combining image recognition and report editing
CN111161290B (zh) 一种图像分割模型的构建方法、图像分割方法及系统
CN107909622B (zh) 模型生成方法、医学成像的扫描规划方法及医学成像系统
CN108416776A (zh) 图像识别方法、图像识别装置、计算机产品和可读存储介质
CN107564048B (zh) 基于分叉点特征配准方法
US11783488B2 (en) Method and device of extracting label in medical image
WO2021208739A1 (zh) 眼底彩照图像血管评估方法、装置、计算机设备和介质
EP3910592A1 (en) Image matching method, apparatus and device, and storage medium
CN111984771A (zh) 一种基于智能对话的自动问诊系统
CN108520512B (zh) 一种测量眼部参数的方法及装置
CN111161268B (zh) 图像处理方法、装置、电子设备及计算机存储介质
JP2013200590A (ja) 類似画像検索装置、方法及びプログラム
CN108510493A (zh) 医学图像内目标对象的边界定位方法、存储介质及终端
CN114627067A (zh) 一种基于图像处理的伤口面积测量及辅助诊疗方法
Salih et al. Fast optic disc segmentation using FFT-based template-matching and region-growing techniques
CN112950737A (zh) 基于深度学习的眼底荧光造影图像生成方法
CN111428070A (zh) 眼科案例的检索方法、装置、服务器及存储介质
CN117877691B (zh) 一种基于图像识别的智能化伤口信息采集系统
CN105869151B (zh) 舌分割及舌苔舌质分离方法
WO2022077858A1 (zh) 基于深度学习体征结构的全模态医学影像序列分组方法
CN104809425B (zh) 手部兴趣区域的提取方法及装置
CN108665474A (zh) 一种基于b-cosfire的眼底图像视网膜血管分割方法
WO2020118826A1 (zh) 一种左心室图像分割方法、装置、设备及存储介质

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: 18931497

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 30.06.2021)

122 Ep: pct application non-entry in european phase

Ref document number: 18931497

Country of ref document: EP

Kind code of ref document: A1