CN112149688A - Image processing method and device, computer readable storage medium, computer device - Google Patents

Image processing method and device, computer readable storage medium, computer device Download PDF

Info

Publication number
CN112149688A
CN112149688A CN202011019693.0A CN202011019693A CN112149688A CN 112149688 A CN112149688 A CN 112149688A CN 202011019693 A CN202011019693 A CN 202011019693A CN 112149688 A CN112149688 A CN 112149688A
Authority
CN
China
Prior art keywords
image
target
pixel
level
image processing
Prior art date
Legal status (The legal status 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 status listed.)
Pending
Application number
CN202011019693.0A
Other languages
Chinese (zh)
Inventor
由佳
孟祥雨
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Automotive Research Institute Co Ltd
Original Assignee
Beijing Automotive Research Institute Co Ltd
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 Beijing Automotive Research Institute Co Ltd filed Critical Beijing Automotive Research Institute Co Ltd
Priority to CN202011019693.0A priority Critical patent/CN112149688A/en
Publication of CN112149688A publication Critical patent/CN112149688A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V2201/00—Indexing scheme relating to image or video recognition or understanding
    • G06V2201/07—Target detection

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses an image processing method and device, a computer readable storage medium and computer equipment, wherein the image processing method comprises the following steps: processing the input image by adopting a target image processing algorithm to obtain a pixel-level target image, and generating a super-pixel-level target image according to the pixel-level target image; performing image segmentation processing on the target image at the super-pixel level to respectively extract a background sample and a binary image; carrying out GOP region prediction on the binarized image to generate an initial saliency map, and extracting a foreground sample according to the initial saliency map; and carrying out model training according to the background sample and the foreground sample. Therefore, the image processing method can ensure the accuracy of the target image training sample, thereby improving the accuracy of target image detection.

Description

Image processing method and device, computer readable storage medium, computer device
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to an image processing method, a computer-readable storage medium, a computer device, and an image processing apparatus.
Background
Image saliency detection has attracted much attention in recent years, and as a preprocessing method, image saliency detection is widely used in a plurality of fields such as image compression, image classification, and image segmentation. In the related technology, the significance detection research mainly utilizes manual design features and heuristic prior to detect significance regions in images, wherein one scheme is that a foreground and background training sample is selected from one image, the foreground training sample selects the center position of the image, the background training sample selects pixels around the image, and if a detection target is not located in the center region of the image, the selected foreground training sample is inaccurate; the other scheme is that a foreground training sample is selected from an initial saliency map, the foreground training sample selects a part with a larger threshold value in the saliency map, and a background training sample selects a part with a smaller threshold value in the saliency map.
Disclosure of Invention
The present invention is directed to solving, at least to some extent, one of the technical problems in the related art. Therefore, an object of the present invention is to provide an image processing method, which can ensure the accuracy of a target image training sample, thereby improving the accuracy of target image detection.
A second object of the invention is to propose a computer-readable storage medium.
A third object of the invention is to propose a computer device.
A fourth object of the present invention is to provide an image processing apparatus.
To achieve the above object, an embodiment of a first aspect of the present invention provides an image processing method, including: processing an input image by adopting a target image processing algorithm to obtain a pixel-level target image, and generating a super-pixel-level target image according to the pixel-level target image; performing image segmentation processing on the target image at the super pixel level to respectively extract a background sample and a binary image; performing GOP (Group of Picture) regional prediction on the binarized image to generate an initial saliency map, and extracting foreground samples according to the initial saliency map; and performing model training according to the background sample and the foreground sample.
The image processing method of the embodiment of the invention firstly adopts a target image processing algorithm to process an input image and obtain a pixel-level target image, then generates a super-pixel-level target image according to the pixel-level target image, then performs image segmentation processing on the super-pixel-level target image to respectively extract a background sample and a binary image, then performs GOP region prediction on the binary image to generate an initial saliency map, then extracts a foreground sample according to the initial saliency map, and finally performs model training according to the background sample and the foreground sample. Therefore, the image processing method can ensure the accuracy of the target image training sample, thereby improving the accuracy of target image detection.
In some examples of the present invention, the pixel-level target map is a plurality of target maps, wherein generating the superpixel-level target map from the pixel-level target map includes: and clustering the target graphs of a plurality of pixel levels to obtain the target graph of the superpixel level.
In some examples of the invention, GOP region prediction is performed on the binarized image to generate an initial saliency map, including: and taking the binarization result of the target map at the super-pixel level as an assumed true value, calculating the F-measure value of each prediction region, and calculating the initial saliency map according to the pixels of the prediction regions of which the F-measure values are greater than a preset value.
In some examples of the invention, the initial saliency map is calculated according to the following formula:
Figure BDA0002698974440000021
where j is the index of the M prediction regions,
Figure BDA0002698974440000022
represents the jth prediction region
Figure BDA0002698974440000023
F-measure value of (1).
To achieve the above object, a second aspect of the present invention provides a computer-readable storage medium, on which an image processing program is stored, the image processing program, when executed by a processor, implementing the image processing method in the above embodiments.
The computer-readable storage medium of the embodiment of the present invention can implement the image processing method in the above-described embodiment by executing the image processing program stored thereon, so as to ensure the accuracy of the target image training sample, thereby improving the accuracy of target image detection.
To achieve the above object, a third embodiment of the present invention provides an apparatus, which includes a memory, a processor, and an image processing program stored in the memory and executable on the processor, and when the processor executes the image processing program, the apparatus implements the image processing method according to the above embodiment.
The device comprises a memory, a processor and an image processing program which is stored on the memory and can be operated on the processor, and the image processing method in the embodiment can be realized by executing the image processing program, so that the accuracy of a target image training sample can be ensured, and the accuracy of target image detection can be improved.
To achieve the above object, a fourth aspect of the present invention provides an image processing apparatus, including: the target image generation module is used for processing the input image by adopting a target image processing algorithm to obtain a pixel-level target image and generating a super-pixel-level target image according to the pixel-level target image; the image segmentation processing module is used for carrying out image segmentation processing on the target image at the superpixel level so as to respectively extract a background sample and a binary image; the extraction module is used for predicting a GOP region of the binarized image to generate an initial saliency map and extracting a foreground sample according to the initial saliency map; and the training module is used for carrying out model training according to the background sample and the foreground sample.
The image processing device comprises a target image generation module, an image segmentation processing module, an extraction module and a training module. Firstly, processing an input image by a target image processing algorithm by using a target image generation module to obtain a pixel-level target image, generating a super-pixel-level target image according to the pixel-level target image, performing image segmentation processing on the super-pixel-level target image by using an image segmentation processing module to respectively extract a background sample and a binary image, performing GOP (group of picture) region prediction on the binary image by using an extraction module to generate a dehumidification saliency map, extracting a foreground sample according to the initial saliency map, and finally performing model training according to the background sample and the foreground sample by using a training module. Therefore, the image processing device can ensure the accuracy of the target image training sample, and the accuracy of target image detection is improved.
In some examples of the present invention, the pixel-level target map is a plurality of target maps, and the target map generation module is further configured to perform a clustering process on the plurality of pixel-level target maps to obtain the superpixel-level target map.
In some examples of the present invention, the extraction module is further configured to calculate an F-measure value of each prediction region using a binarization result of the target map at the superpixel level as an assumed true value, and calculate the initial saliency map from pixels of the prediction region having the F-measure value greater than a preset value.
In some examples of the invention, the initial saliency map is calculated according to the following formula:
Figure BDA0002698974440000031
where j is the index of the M prediction regions,
Figure BDA0002698974440000032
represents the jth prediction region
Figure BDA0002698974440000033
F-measure value of (1).
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
FIG. 1 is a flow chart of an image processing method of an embodiment of the present invention;
FIG. 2 is a schematic diagram of an image processing process according to an embodiment of the invention;
fig. 3 is a block diagram of the image processing apparatus according to the embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
An image processing method and apparatus, a computer-readable storage medium, and a computer device according to embodiments of the present invention are described below with reference to the accompanying drawings.
Fig. 1 is a flowchart of an image processing method of an embodiment of the present invention.
In this embodiment, as shown in fig. 1, the image processing method includes the steps of:
s10, processing the input image using a target image processing algorithm to obtain a target map at the pixel level, and generating a target map at the superpixel level from the target map at the pixel level.
Specifically, referring to fig. 2, wherein fig. 2(a) is an input image, in this embodiment, the input image may be processed by a target image processing algorithm to obtain a target map at the pixel level as shown in fig. 2 (b). In some examples of the present invention, as shown in fig. 2(b), the target map at the pixel level may be plural, wherein the generating of the target map at the super-pixel level from the target map at the pixel level includes: and clustering the target graphs of the multiple pixel levels to obtain a target graph of a super pixel level.
It should be noted that in this embodiment, an Objectness algorithm may be used to calculate a target score for the target map at each pixel level, and as shown in fig. 2(b), the target scores may be respectively denoted as S1, S2, and the like, and are weights of the target maps at their respective pixel levels. After the target score of the target map of each pixel level is obtained through calculation, the target maps of the pixel levels with the target scores are subjected to clustering processing through a target map clustering method to obtain a target map of a super pixel level, namely, a map 2 (c). Specifically, the target maps at the respective pixel levels are multiplied by their respective target scores and then summed to obtain the target map at the superpixel level.
In some examples of the invention, the target image processing algorithm may be an Objectness algorithm, i.e. the input image is processed by the Objectness algorithm to obtain the target map at pixel level.
S20, a super-pixel level target map is subjected to image segmentation processing to extract a background sample and a binarized image, respectively.
Specifically, after the target maps at multiple pixel levels are subjected to clustering processing to obtain the target maps at the superpixel level, image segmentation processing may be performed on the target maps at the superpixel level again, more specifically, binarization processing is performed on the target maps at the superpixel level to obtain a binarized image, and in this embodiment, fig. 2(d) is recorded as the binarized image obtained after image segmentation processing. It should be noted that, after the target map at the super-pixel level is subjected to the graph cutting process, a background sample of the input image may also be extracted.
And S30, performing GOP region prediction on the binary image to generate an initial saliency map, and extracting foreground samples according to the initial saliency map.
After the image segmentation process and the binarized image are obtained, GOP region prediction may be performed on the binarized image to generate an initial saliency map. In some examples, an F-measure value of each prediction region may be calculated using a binarization result of the target map at a super-pixel level as an assumed true value, and the initial saliency map may be calculated from pixels of the prediction region having the F-measure value greater than a preset value. Alternatively, the initial saliency map may be calculated according to the following formula:
Figure BDA0002698974440000041
where j is the index of the M prediction regions,
Figure BDA0002698974440000042
represents the jth prediction region
Figure BDA0002698974440000043
F-measure value of (1).
Specifically, in this embodiment, fig. 2(e) can be obtained after taking the binarization result as an assumed true value and calculating the F-measure value of each prediction region. In FIG. 2(e), Fm1 represents the F-measure value of the first prediction region, Fm2 represents the F-measure value of the second prediction region, and so on. It should be noted that the F-measure value is a commonly used evaluation criterion in the field of information retrieval, and is commonly used to evaluate the quality of a classification model, and in different models, different F-measure values can be selected to find a balance point between the recall rate and the accuracy rate in the model. In this embodiment, the initial saliency map, fig. 2(F), can be generated by linearly weighting the predicted regions corresponding to the F-measure values greater than the preset value or the first 40% of the F-measure values.
More specifically, in this embodiment, the expression
Figure BDA0002698974440000051
Calculating an initial saliency map, wherein i represents the ith pixel of the target image, and the pixel i can be in M prediction regions, so that S of i in each M prediction region can be obtainedin(i) Value, then weighted, Sin(i) Representing the probability that the current pixel i is a foreground pixel,
Figure BDA0002698974440000052
used for judging whether the current pixel i is a foreground pixel in the jth index, if so, the current pixel i is a foreground pixel
Figure BDA0002698974440000053
Has a value of 1, i.e. the current value needs to be taken into account
Figure BDA0002698974440000059
A value; and if the current pixel i is the background pixel in the jth index
Figure BDA0002698974440000054
Is 0, i.e. without taking into account the current background pixel
Figure BDA0002698974440000055
The value is obtained. It is understood that if
Figure BDA0002698974440000056
The larger the probability that the current pixel is a foreground pixel, Sin(i) The larger the value of (A), the S of all pixels in the calculated imageinAfter the values, all pixels can then be composed into one foreground sample. It should be noted that, when all pixels are combined into a foreground sample, some background pixels may be miscalculated, but the background pixels are not miscalculated
Figure BDA0002698974440000057
Of values much smaller than the foreground pixels
Figure BDA0002698974440000058
The values, even if a portion of the background pixels are miscalculated, are displayed in the resulting foreground sample as being very dark and negligible, while the foreground pixels are highlighted as being very bright. After the initial saliency map is obtained through the formula calculation, the generated initial saliency map can better highlight the foreground, and the foreground sample can be extracted according to the initial saliency map.
And S40, performing model training according to the background sample and the foreground sample.
Specifically, after the initial saliency map is obtained through calculation, since the initial saliency map can well highlight the foreground, the foreground sample can be selected by using the initial saliency map, so that some background regions can be prevented from being selected by mistake, and the accuracy of the foreground sample is ensured. And the background sample extracted after the image segmentation processing is carried out on the target image of the super-pixel level can well inhibit the foreground area under the control of the threshold value, so that the background sample is more accurate. After accurate background samples and foreground samples are obtained, model training can be carried out, and therefore accuracy of model training samples can be better guaranteed.
In conclusion, the image processing method provided by the embodiment of the invention can ensure the accuracy of the target image training sample, thereby improving the accuracy of target image detection.
Further, the present invention proposes a computer-readable storage medium on which an image processing program is stored, which when executed by a processor implements the image processing method as in the above-described embodiments.
The computer-readable storage medium of the embodiment of the invention can ensure the accuracy of the target image training sample when the image processing program stored thereon and corresponding to the image processing method is executed by the processor, thereby improving the accuracy of target image detection.
Further, the present invention proposes a computer device comprising a memory, a processor and an image processing program stored on the memory and executable on the processor, the processor implementing the image processing method as in the above embodiments when executing the image processing program.
The device in the embodiment of the invention comprises a memory and a processor, and when an image processing program stored on the memory and corresponding to the image processing method in the embodiment is executed by the processor, the accuracy of the target image training sample can be ensured, so that the accuracy of target image detection is improved.
Fig. 3 is a block diagram of the image processing apparatus according to the embodiment of the present invention.
Further, the present invention proposes an image processing apparatus 100, as shown in fig. 3, the image processing apparatus 100 includes a target graph generation module 101, a graph cut processing module 102, an extraction module 103, and a training module 104.
The target image generation module 101 is configured to process an input image by using a target image processing algorithm to obtain a pixel-level target image, and generate a super-pixel-level target image according to the pixel-level target image; the image segmentation processing module 102 is configured to perform image segmentation processing on the target image at the superpixel level to extract a background sample and a binarized image respectively; the extraction module 103 is configured to perform GOP region prediction on the binarized image to generate an initial saliency map, and extract a foreground sample according to the initial saliency map; the training module 104 is configured to perform model training according to the background sample and the foreground sample.
Specifically, referring to fig. 2, where fig. 2(a) is an input image, in this embodiment, the input image may be processed by the target generation module 101 using a target image processing algorithm to obtain a target map at the pixel level as shown in fig. 2 (b). In some examples of the present invention, as shown in fig. 2(b), the number of the pixel-level target maps may be multiple, and the target map generation module 101 is further configured to perform a clustering process on the multiple pixel-level target maps to obtain the superpixel-level target map.
It should be noted that in this embodiment, an Objectness algorithm may be used to calculate a target score for the target map at each pixel level, and as shown in fig. 2(b), the target scores may be respectively denoted as S1, S2, and the like, and are weights of the target maps at their respective pixel levels. After the target score of the target map of each pixel level is obtained through calculation, the target maps of the pixel levels with the target scores are subjected to clustering processing through a target map clustering method to obtain a target map of a super pixel level, namely, a map 2 (c). Specifically, the target maps at the respective pixel levels are multiplied by their respective target scores and then summed to obtain the target map at the superpixel level.
In some examples of the invention, the target image processing algorithm may be an Objectness algorithm, i.e. the input image is processed by the Objectness algorithm to obtain the target map at pixel level.
After the target map at the superpixel level is obtained, the target map at the superpixel level may be subjected to image segmentation processing by the image segmentation processing module 102 to extract the background sample and the binarized image, respectively.
Specifically, after the target map generation module 101 performs clustering processing on the target maps at multiple pixel levels and obtains the target maps at the superpixel level, the graph cut processing module 102 may be used to perform graph cut processing on the target maps at the superpixel level again, more specifically, perform binarization processing on the target maps at the superpixel level to obtain a binarized image, and in this embodiment, fig. 2(d) is recorded as the binarized image obtained after the graph cut processing. It should be noted that, after the target map at the super-pixel level is subjected to the graph cutting process, a background sample of the input image may also be extracted.
After the binarized image is obtained, the extraction module 103 is used for performing GOP region prediction on the binarized image to generate an initial saliency map, and a foreground sample is extracted according to the initial saliency map.
Specifically, after the target map at the super-pixel level is subjected to the image segmentation processing by the image segmentation processing module 102 and a binarized image is obtained, the binarized image may be subjected to GOP region prediction by the extraction module 103 to generate an initial saliency map. In some examples, the extraction module 103 may calculate an F-measure value of each prediction region using a binarization result of the target map at a super-pixel level as an assumed true value, and calculate an initial saliency map from pixels of the prediction region having the F-measure value greater than a preset value. Alternatively, the initial saliency map may be calculated according to the following formula:
Figure BDA0002698974440000071
where j is the index of the M prediction regions,
Figure BDA0002698974440000072
represents the jth prediction region
Figure BDA0002698974440000073
F-measure value of (1).
Specifically, in this embodiment, fig. 2(e) can be obtained after the extraction module 103 uses the binarization result as an assumed true value and calculates the F-measure value of each prediction region. In FIG. 2(e), Fm1 represents the F-measure value of the first prediction region, Fm2 represents the F-measure value of the second prediction region, and so on. It should be noted that the F-measure value is a commonly used evaluation criterion in the field of information retrieval, and is commonly used to evaluate the quality of a classification model, and in different models, different F-measure values can be selected to find a balance point between the recall rate and the accuracy rate in the model. In this embodiment, the prediction regions corresponding to the F-measure values larger than the predetermined value or the first 40% of the F-measure values may be selected for linear weighting to generate the initial saliency map, which is shown in fig. 2 (F).
More specifically, in this embodiment, the expression
Figure BDA0002698974440000074
Calculating an initial saliency map, wherein i represents the ith pixel of the target image, and the pixel i can be in M prediction regions, so that S of i in each M prediction region can be obtainedin(i) Value, then weighted, Sin(i) Representing the probability that the current pixel i is a foreground pixel,
Figure BDA0002698974440000075
used for judging whether the current pixel i is a foreground pixel in the jth index, if so, the current pixel i is a foreground pixel
Figure BDA0002698974440000076
Has a value of 1, i.e. the current value needs to be taken into account
Figure BDA0002698974440000077
A value; and if the current pixel i is the background pixel in the jth index
Figure BDA0002698974440000078
Is 0, i.e. without taking into account the current background pixel
Figure BDA0002698974440000079
The value is obtained. It is understood that if
Figure BDA00026989744400000710
The larger the probability that the current pixel is a foreground pixel, Sin(i) The larger the value of (A), the S of all pixels in the calculated imageinAfter the values, all pixels can then be composed into one foreground sample. It should be noted that, when all pixels are combined into a foreground sample, some background pixels may be miscalculated thereinBut due to background pixels
Figure BDA00026989744400000711
Of values much smaller than the foreground pixels
Figure BDA00026989744400000712
The values, even if a portion of the background pixels are miscalculated, are displayed in the resulting foreground sample as being very dark and negligible, while the foreground pixels are highlighted as being very bright. After the initial saliency map is obtained through the formula calculation, the generated initial saliency map can better highlight the foreground, and the foreground sample can be extracted according to the initial saliency map.
After the foreground sample and the background sample are obtained, the training module 104 is utilized to perform model training according to the background sample and the foreground sample.
Specifically, after the initial saliency map is obtained through calculation, since the initial saliency map can well highlight the foreground, the foreground sample can be selected by using the initial saliency map, so that some background regions can be prevented from being selected by mistake, and the accuracy of the foreground sample is ensured. And the background sample extracted after the image segmentation processing is carried out on the target image of the super-pixel level can well inhibit the foreground area under the control of the threshold value, so that the background sample is more accurate. After accurate background samples and foreground samples are obtained, model training can be performed by using the training module 104, so that the accuracy of model training samples can be better ensured.
In summary, the image processing device of the embodiment of the invention can ensure the accuracy of the target image training sample, thereby improving the accuracy of target image detection.
It should be noted that the logic and/or steps represented in the flowcharts or otherwise described herein, such as an ordered listing of executable instructions that can be considered to implement logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium would include the following: an electrical connection (electronic device) having one or more wires, a portable computer diskette (magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium could even be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via for instance optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner if necessary, and then stored in a computer memory.
It should be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the various steps or methods may be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or combination of the following techniques, which are known in the art, may be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application specific integrated circuit having an appropriate combinational logic gate circuit, a Programmable Gate Array (PGA), a Field Programmable Gate Array (FPGA), or the like.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
In the description of the present invention, it is to be understood that the terms "central," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," "circumferential," and the like are used in the orientations and positional relationships indicated in the drawings for convenience in describing the invention and to simplify the description, and are not intended to indicate or imply that the referenced devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and are therefore not to be considered limiting of the invention.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
In the present invention, unless otherwise expressly stated or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; they may be directly connected or indirectly connected through intervening media, or they may be connected internally or in any other suitable relationship, unless expressly stated otherwise. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
In the present invention, unless otherwise expressly stated or limited, the first feature "on" or "under" the second feature may be directly contacting the first and second features or indirectly contacting the first and second features through an intermediate. Also, a first feature "on," "over," and "above" a second feature may be directly or diagonally above the second feature, or may simply indicate that the first feature is at a higher level than the second feature. A first feature being "under," "below," and "beneath" a second feature may be directly under or obliquely under the first feature, or may simply mean that the first feature is at a lesser elevation than the second feature.
Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention, and that variations, modifications, substitutions and alterations can be made to the above embodiments by those of ordinary skill in the art within the scope of the present invention.

Claims (10)

1. An image processing method, comprising:
processing an input image by adopting a target image processing algorithm to obtain a pixel-level target image, and generating a super-pixel-level target image according to the pixel-level target image;
performing image segmentation processing on the target image at the super pixel level to respectively extract a background sample and a binary image;
carrying out GOP region prediction on the binarized image to generate an initial saliency map, and extracting a foreground sample according to the initial saliency map;
and performing model training according to the background sample and the foreground sample.
2. The image processing method according to claim 1, wherein the plurality of pixel-level target maps are plural, and wherein generating the superpixel-level target map from the pixel-level target maps comprises:
and clustering the target graphs of a plurality of pixel levels to obtain the target graph of the superpixel level.
3. The image processing method according to any one of claims 1-2, wherein performing GOP region prediction on the binarized image to generate an initial saliency map comprises:
and taking the binarization result of the target map at the super-pixel level as an assumed true value, calculating the F-measure value of each prediction region, and calculating the initial saliency map according to the pixels of the prediction regions of which the F-measure values are greater than a preset value.
4. An image processing method according to claim 3, characterized in that the initial saliency map is calculated according to the following formula:
Figure FDA0002698974430000011
where j is the index of the M prediction regions,
Figure FDA0002698974430000012
represents the jth prediction region
Figure FDA0002698974430000013
F-measure value of (1).
5. A computer-readable storage medium, having stored thereon an image processing program which, when executed by a processor, implements the image processing method according to any one of claims 1 to 4.
6. A computer device comprising a memory, a processor and an image processing program stored on the memory and executable on the processor, the processor implementing the image processing method as claimed in any one of claims 1 to 4 when executing the image processing program.
7. An image processing apparatus characterized by comprising:
the target image generation module is used for processing the input image by adopting a target image processing algorithm to obtain a pixel-level target image and generating a super-pixel-level target image according to the pixel-level target image;
the image segmentation processing module is used for carrying out image segmentation processing on the target image at the superpixel level so as to respectively extract a background sample and a binary image;
the extraction module is used for predicting a GOP region of the binarized image to generate an initial saliency map and extracting a foreground sample according to the initial saliency map;
and the training module is used for carrying out model training according to the background sample and the foreground sample.
8. The image processing apparatus according to claim 7, wherein the pixel-level target map is a plurality of target maps, and wherein the target map generation module is further configured to perform a clustering process on the plurality of pixel-level target maps to obtain the superpixel-level target map.
9. The image processing apparatus according to claim 7 or 8, wherein the extraction module is further configured to calculate an F-measure value of each prediction region using a binarization result of the target map at the superpixel level as an assumed true value, and to calculate the initial saliency map from pixels of prediction regions whose F-measure values are greater than a preset value.
10. The image processing apparatus according to claim 9, wherein the initial saliency map is calculated according to the following formula:
Figure FDA0002698974430000021
where j is the index of the M prediction regions,
Figure FDA0002698974430000022
represents the jth prediction region
Figure FDA0002698974430000024
F-measure value of (1).
CN202011019693.0A 2020-09-24 2020-09-24 Image processing method and device, computer readable storage medium, computer device Pending CN112149688A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011019693.0A CN112149688A (en) 2020-09-24 2020-09-24 Image processing method and device, computer readable storage medium, computer device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011019693.0A CN112149688A (en) 2020-09-24 2020-09-24 Image processing method and device, computer readable storage medium, computer device

Publications (1)

Publication Number Publication Date
CN112149688A true CN112149688A (en) 2020-12-29

Family

ID=73897929

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011019693.0A Pending CN112149688A (en) 2020-09-24 2020-09-24 Image processing method and device, computer readable storage medium, computer device

Country Status (1)

Country Link
CN (1) CN112149688A (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150262039A1 (en) * 2014-03-13 2015-09-17 Omron Corporation Image processing apparatus and image processing method
CN106960434A (en) * 2017-03-03 2017-07-18 大连理工大学 A kind of image significance detection method based on surroundedness and Bayesian model
CN107169487A (en) * 2017-04-19 2017-09-15 西安电子科技大学 The conspicuousness object detection method positioned based on super-pixel segmentation and depth characteristic
CN110443257A (en) * 2019-07-08 2019-11-12 大连理工大学 A kind of conspicuousness detection method based on Active Learning

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150262039A1 (en) * 2014-03-13 2015-09-17 Omron Corporation Image processing apparatus and image processing method
CN106960434A (en) * 2017-03-03 2017-07-18 大连理工大学 A kind of image significance detection method based on surroundedness and Bayesian model
CN107169487A (en) * 2017-04-19 2017-09-15 西安电子科技大学 The conspicuousness object detection method positioned based on super-pixel segmentation and depth characteristic
CN110443257A (en) * 2019-07-08 2019-11-12 大连理工大学 A kind of conspicuousness detection method based on Active Learning

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
JIA YOU, 等: "Salient object detection via point-to-set metric learning", 《PATTERN RECOGNITION LETTERS》, 31 August 2016 (2016-08-31), pages 85 - 90 *
由佳: "基于点—集度量学习的显著目标检测", 《中国优秀硕士学位论文全文数据库 信息科技辑》, 15 March 2017 (2017-03-15), pages 138 - 4549 *

Similar Documents

Publication Publication Date Title
EP3101594A1 (en) Saliency information acquisition device and saliency information acquisition method
CN111524137B (en) Cell identification counting method and device based on image identification and computer equipment
CN109948683B (en) Method and device for dividing difficulty degree of point cloud data and related equipment thereof
CN112329702B (en) Method and device for rapid face density prediction and face detection, electronic equipment and storage medium
EP3039645B1 (en) A semi automatic target initialization method based on visual saliency
CN112132131B (en) Measuring cylinder liquid level identification method and device
US20170178341A1 (en) Single Parameter Segmentation of Images
CN114549445A (en) Image detection and related model training method, related device, equipment and medium
CN115375917B (en) Target edge feature extraction method, device, terminal and storage medium
CN116030237A (en) Industrial defect detection method and device, electronic equipment and storage medium
US20170053172A1 (en) Image processing apparatus, and image processing method
CN114155398A (en) Label type self-adaptive active learning image target detection method and device
JP2011165170A (en) Object detection device and program
EP2902969B1 (en) Image processing device, image processing method, and image processing program
CN111062954B (en) Infrared image segmentation method, device and equipment based on difference information statistics
CN115908999B (en) Method for detecting rust of top hardware fitting of distribution pole tower, medium and edge terminal equipment
CN112149688A (en) Image processing method and device, computer readable storage medium, computer device
CN116486153A (en) Image classification method, device, equipment and storage medium
CN114445716B (en) Key point detection method, key point detection device, computer device, medium, and program product
CN113537253B (en) Infrared image target detection method, device, computing equipment and storage medium
CN111723719B (en) Video target detection method, system and device based on category external memory
CN111914766B (en) Method for detecting business trip behavior of city management service
CN113326749A (en) Target detection method and device, storage medium and electronic equipment
CN115497172B (en) Fishing behavior detection method and device, edge processing equipment and storage medium
CN111275045A (en) Method and device for identifying image subject, electronic equipment and medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination