CN117934909A - Identification object candidate region screening method and device, electronic equipment and storage medium - Google Patents

Identification object candidate region screening method and device, electronic equipment and storage medium Download PDF

Info

Publication number
CN117934909A
CN117934909A CN202311775066.3A CN202311775066A CN117934909A CN 117934909 A CN117934909 A CN 117934909A CN 202311775066 A CN202311775066 A CN 202311775066A CN 117934909 A CN117934909 A CN 117934909A
Authority
CN
China
Prior art keywords
candidate
local binary
candidate frame
texture
entropy
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
CN202311775066.3A
Other languages
Chinese (zh)
Inventor
张砚
陈瑞
叶齐祥
贾惠柱
李源
葛盛
辛煜
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 Institute of Remote Sensing Information
Original Assignee
Beijing Institute of Remote Sensing Information
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 Institute of Remote Sensing Information filed Critical Beijing Institute of Remote Sensing Information
Priority to CN202311775066.3A priority Critical patent/CN117934909A/en
Publication of CN117934909A publication Critical patent/CN117934909A/en
Pending legal-status Critical Current

Links

Classifications

    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Image Analysis (AREA)

Abstract

The invention provides a screening method, a screening device, electronic equipment and a storage medium for candidate areas of identification objects, wherein the number of complete contours included in a candidate frame is obtained according to a candidate frame edge diagram by obtaining the candidate frame edge diagram of an image to be detected; obtaining a texture characteristic value of an image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object; texture complexity scores are calculated according to the number of complete contours and the entropy of the local binary patterns included in the candidate frames, the texture complexity scores are used as confidence levels for measuring each candidate region, candidate regions of the identification objects are screened out according to the confidence levels, redundancy of the candidate regions can be reduced, and the efficiency and the accuracy of remote sensing image detection are improved.

Description

Identification object candidate region screening method and device, electronic equipment and storage medium
Technical Field
The present invention relates to the field of image processing technologies, and in particular, to a method and apparatus for screening candidate regions of an identification object, an electronic device, and a storage medium.
Background
In the existing remote sensing image detection process, in order to identify a candidate region where an identification object is located, an image segmentation and image merging process is performed for a given image using an image segmentation method and a hierarchical region merging method to acquire the candidate region. Performing the image segmentation and merging process results in the generation of redundant regions, which are generated because the image segmentation process first initializes the image to uniformly colored pixel blocks, i.e., super-pixels, using an image wind algorithm, and then performs a merging operation, which releases the current region when every two blobs are merged from the hierarchical merge structure until multiple segmented blobs merge into one region. To achieve high recall, multiple color spaces, different similarity metric combinations, and initialized pixel block sizes are used, which, while increasing the number of windows, makes it possible to avoid losing any recognition objects by candidate windows, creates redundant regions. For natural scene images, tens of thousands of candidate regions are generated, many of these candidate regions are redundant regions, the windows of the redundant regions are highly overlapped with each other, and one object may be covered multiple times, which affects the efficiency and accuracy of remote sensing image detection.
Disclosure of Invention
The invention provides a screening method, a screening device, electronic equipment and a storage medium for candidate areas of identification objects, which are used for solving the defect that redundant areas exist in the candidate areas where the traditional identification objects are located, and the efficiency and the accuracy of remote sensing image detection are affected.
The invention provides a screening method of candidate areas of identification objects, which comprises the following steps:
Acquiring a candidate frame edge map of an image to be detected, and acquiring the number of complete contours included in a candidate frame according to the candidate frame edge map;
Obtaining a texture characteristic value of the image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object;
And calculating texture complexity scores according to the number of complete contours included in the candidate frames and the local binary pattern entropy, taking the texture complexity scores as confidence measures of each candidate region, and screening out candidate regions of the identification object according to the confidence measures.
According to the method for screening the candidate region of the identification object provided by the invention, the method for acquiring the candidate frame edge map of the image to be detected comprises the following steps:
And acquiring the edge intensity and the direction of each pixel point in the image to be detected based on a method of a structure forest, and determining an edge map of each candidate frame according to the edge intensity and the direction of each pixel point.
According to the method for screening the candidate areas of the identification objects provided by the invention, the method for acquiring the number of complete contours included in the candidate frames according to the edge map of the candidate frames comprises the following steps:
Acquiring the sum of all edge map areas in each candidate frame, the width and the height of each candidate frame, and a weight function corresponding to the complete outline in each candidate frame;
and calculating the number of complete contours included in each candidate frame according to the sum of all edge map areas in each candidate frame, the width and the height of each candidate frame and the weight function.
According to the identification object candidate region screening method provided by the invention, the weight function is calculated according to the affinity and the length of each contour in the average direction.
According to the method for screening the candidate region of the identification object provided by the invention, the obtaining of the texture characteristic value of the image to be detected comprises the following steps:
In a3×3 window, using the gray value of the pixel point at the center of the window as a threshold value, comparing the gray value of 8 adjacent pixels with the threshold value, if the gray value is larger than the threshold value, marking the position of the corresponding pixel point as 1, otherwise, setting the position as 0, generating 8-bit binary numbers, and taking the 8-bit binary numbers as texture characteristic values of the pixel point at the center of the window.
According to the method for screening the candidate region of the identification object provided by the invention, the local binary pattern entropy is calculated according to the texture characteristic value, and the method comprises the following steps:
Calculating the probability of each mode corresponding to the value of the local binary mode according to the value of the local binary mode and the width and the height of the candidate region;
and calculating the local binary mode entropy according to the probability of each mode.
According to the method for screening the candidate regions of the identification objects provided by the invention, the texture complexity score is calculated according to the number of complete contours included in the candidate frames and the local binary pattern entropy, and the method comprises the following steps:
Designing a gate function of the local binary pattern entropy, wherein the gate function comprises a plurality of extraction thresholds;
And constructing a texture complexity calculation formula according to the gate function and the number of the complete contours, extracting candidate frames meeting the proper area according to an extraction threshold, and calculating the texture complexity score of the candidate region.
The invention also provides a screening device for the candidate areas of the identification objects, which comprises the following steps:
The first acquisition module is used for acquiring a candidate frame edge map of the image to be detected, and acquiring the number of complete contours included in the candidate frame according to the candidate frame edge map;
the second acquisition module is used for acquiring the texture characteristic value of the image to be detected, calculating a local binary pattern entropy according to the texture characteristic value, and the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object;
and the screening module is used for calculating texture complexity scores according to the number of the complete contours included in the candidate frames and the local binary pattern entropy, using the texture complexity scores as confidence measures for measuring each candidate region, and screening the candidate regions of the identification objects according to the confidence measures.
The invention also provides electronic equipment, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor realizes the candidate area screening method of the identification object when executing the program.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the recognition object candidate region screening method of any one of the above.
According to the identification object candidate region screening method, the identification object candidate region screening device, the electronic equipment and the storage medium, the number of complete contours included in a candidate frame is obtained according to the candidate frame edge map by obtaining the candidate frame edge map of an image to be detected; obtaining a texture characteristic value of an image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object; texture complexity scores are calculated according to the number of complete contours and the entropy of the local binary patterns included in the candidate frames, the texture complexity scores are used as confidence levels for measuring each candidate region, candidate regions of the identification objects are screened out according to the confidence levels, redundancy of the candidate regions can be reduced, and the efficiency and the accuracy of remote sensing image detection are improved.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a screening method of candidate areas of identification objects provided by the invention;
fig. 2 is a schematic structural diagram of a candidate region screening device for identification objects according to the present invention;
fig. 3 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Fig. 1 is a flowchart of a method for screening candidate regions of an identification object according to an embodiment of the present invention, where, as shown in fig. 1, the method for screening candidate regions of an identification object according to an embodiment of the present invention includes:
Step 101, acquiring a candidate frame edge map of an image to be detected, and acquiring the number of complete contours included in a candidate frame according to the candidate frame edge map;
102, obtaining a texture characteristic value of an image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between an identification object and a non-identification object;
And 103, calculating texture complexity scores according to the number of complete contours and the entropy of the local binary patterns included in the candidate frames, taking the texture complexity scores as confidence degrees for measuring each candidate region, and screening the candidate regions of the identification object according to the confidence degrees.
In the conventional remote sensing image detection process, in order to obtain a high recall rate, multiple color spaces, different similarity measurement combinations and initialized pixel block sizes are used, and in this way, the number of windows can be increased, so that candidate windows cannot lose any recognition objects, but redundant areas are generated. For natural scene images, tens of thousands of candidate regions are generated, many of these candidate regions are redundant regions, the windows of the redundant regions are highly overlapped with each other, and one object may be covered multiple times, which affects the efficiency and accuracy of remote sensing image detection.
According to the identification object candidate region screening method provided by the embodiment of the invention, the number of complete contours included in the candidate frame is obtained according to the candidate frame edge map by obtaining the candidate frame edge map of the image to be detected; obtaining a texture characteristic value of an image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object; texture complexity scores are calculated according to the number of complete contours and the entropy of the local binary patterns included in the candidate frames, the texture complexity scores are used as confidence levels for measuring each candidate region, candidate regions of the identification objects are screened out according to the confidence levels, redundancy of the candidate regions can be reduced, and the efficiency and the accuracy of remote sensing image detection are improved.
Based on any of the above embodiments, the obtaining the candidate frame edge map of the image to be measured in the above steps includes:
and acquiring the edge intensity and the direction of each pixel point in the image to be detected based on a method of the structure forest, and determining the edge map of each candidate frame according to the edge intensity and the direction of each pixel point.
The structure forest based method acquires an edge map of the image, and a complete contour is a collection of adjacent edge points with similar directions. The number of complete contours contained entirely within a candidate window may generally indicate whether or not the window has the presence of a detected object. Definition of the definitionAnd (3) representing an image edge graph with the size W multiplied by H, wherein m n and o n respectively represent the intensity and the direction of the edge of the pixel point, and n is a window mark. If the set of edge groups in an image is s= { S i }, i is the edge group identity, the set of edge groups in the candidate box is/>
In the embodiment of the present invention, obtaining the number of complete contours included in a candidate frame according to a candidate frame edge map includes:
Acquiring the sum of all edge map areas in each candidate frame, the width and the height of each candidate frame, and a weight function corresponding to the complete outline in each candidate frame;
and calculating the number of the complete contours included in each candidate frame according to the sum of all the edge map areas in each candidate frame, the width and the height of each candidate frame and the weight function.
The number of complete contours in the candidate frame is calculated as
Where m i is the sum of the sizes m p of all edges p in the edge group s i, and h b and w b are the width and height of the candidate box. The perimeter of the window, κ, serves as a normalization and κ may serve as a bias to offset the larger window containing more contours.
In the embodiment of the invention, the weight function is calculated according to the affinity and the length of the average direction of each contour.
The weight function f (s i) is defined as:
wherein, Is the affinity of the contours t j and t j+1 in the average direction, and edge P is the ordered path of length |p|. If the contour appears as an overlapping window, the likelihood that it is an object of recognition is reduced.
Based on any of the above embodiments, obtaining the texture feature value of the image to be measured includes:
In a3×3 window, using the gray value of the pixel point at the center of the window as a threshold value, comparing the gray value of 8 adjacent pixels with the threshold value, if the gray value is larger than the threshold value, marking the position of the corresponding pixel point as 1, otherwise, setting the position as 0, generating 8-bit binary numbers, and taking the 8-bit binary numbers as texture characteristic values of the pixel point at the center of the window.
In the embodiment of the present invention, the 8-bit binary number is usually converted into a decimal number, i.e., LBP (Local Binary Pattern ) code, 256 kinds of LBP values are obtained, and the value is used to reflect the texture information of the region.
The calculation formula of the LBP entropy is:
Here p i is the probability of a pattern, the calculation formula is as follows:
Where N i is the number of ith modes, and h b and w b are the width and height of the region, respectively. The LBP entropy of an identified object region is neither too large nor too small.
Calculating the LBP entropy is very efficient because on the preprocessed picture feature map the LBP features of the region are very easy to obtain, while the calculation of the LBP entropy is not scale-proof.
The texture features are represented by using LBP features, so that the local texture features of the image can be well described, and the method has the remarkable advantages of rotation invariance, gray invariance and the like.
The identified object regions typically have dense texture features, whereas the background regions are typically non-textured or diverse. LBP entropy can capture the variability between identified objects and non-identified objects.
In an embodiment of the present invention, calculating a local binary pattern entropy according to the texture feature value includes:
Calculating the probability of each mode corresponding to the value of the local binary mode according to the value of the local binary mode and the width and the height of the candidate region;
and calculating the local binary mode entropy according to the probability of each mode.
Based on any of the above embodiments, computing a texture complexity score from the number of complete contours and the local binary pattern entropy included in the candidate box, comprises:
Designing a gate function of the local binary pattern entropy, wherein the gate function comprises a plurality of extraction thresholds;
On the paspal VOC2007 data training set, the number of complete region contours and LBP entropy were calculated. Identifying object candidates is a trade-off between recall and number of windows. The number of redundant windows is also decreasing with an increase in the threshold number of full contours, but thus the recall of positive samples is decreasing. To achieve a high recall, a small threshold may be required. However, this means a higher false positive rate. The false alarm rate is reduced by combining the LBP entropy and the number of complete contours. The variance of the LBP entropy of the positive example sample is smaller than that of the negative example sample. Thus a gate function of LBP entropy is designed:
Wherein T l、Tml、Tmr and T r are respectively different size profile number thresholds. If the LBP entropy of a window is close to the peak of the distribution, the window is extracted. Meaning that it is more likely to be an object. Candidate boxes with too large or too small LBP entropy values, which are typically non-textured or background areas, are deleted.
And constructing a texture complexity calculation formula according to the gate function and the number of the complete contours, extracting candidate frames meeting the proper area according to the extraction threshold, and calculating the texture complexity score of the candidate region.
The candidate region texture complexity score is defined as
o＝we·g(wt).
The redundant areas are ordered to reduce the number of recognition object candidate areas.
The embodiment of the invention can calculate the score of each window through a formula based on the preprocessed edge graph, and the window description with high score comprises an identification object, which is also a corresponding complete contour with more in the window.
Experiments show that the method provided by the embodiment of the invention has better recall rate when thousands of windows are used. A recall of 75% can be achieved with 655 detection windows and a recall of 25% and 50% can be achieved with only 12 and 91 detection windows, respectively, which is a sufficiently small number of windows in all comparative methods, which has a high performance.
According to the method for screening the candidate areas of the identification objects, provided by the embodiment of the invention, the redundant areas are generated by using a color segmentation method and a layering superpixel merging process, then the texture complexity score is calculated to measure the confidence of each area as an object, the redundancy of the candidate areas can be reduced based on the sorting of the texture complexity score, and the complementarity of the superpixel merging and the object measurement is fully mined.
The recognition object candidate region screening apparatus provided by the present invention will be described below, and the recognition object candidate region screening apparatus described below and the recognition object candidate region screening method described above may be referred to correspondingly to each other.
Fig. 2 is a schematic diagram of an identification object candidate region screening device according to an embodiment of the present invention, where, as shown in fig. 2, the identification object candidate region screening device according to the embodiment of the present invention includes:
A first obtaining module 201, configured to obtain a candidate frame edge map of an image to be detected, and obtain the number of complete contours included in a candidate frame according to the candidate frame edge map;
A second obtaining module 202, configured to obtain a texture feature value of the image to be detected, and calculate a local binary pattern entropy according to the texture feature value, where the local binary pattern entropy is used to characterize a difference between an identified object and a non-identified object;
and the screening module 203 is configured to calculate a texture complexity score according to the number of complete contours included in the candidate frame and the local binary pattern entropy, and use the texture complexity score as a confidence measure for each candidate region, and screen the candidate region of the recognition object according to the confidence measure.
According to the identification object candidate region screening device provided by the embodiment of the invention, the number of complete contours included in a candidate frame is obtained according to the candidate frame edge map by obtaining the candidate frame edge map of an image to be detected; obtaining a texture characteristic value of an image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object; texture complexity scores are calculated according to the number of complete contours and the entropy of the local binary patterns included in the candidate frames, the texture complexity scores are used as confidence levels for measuring each candidate region, candidate regions of the identification objects are screened out according to the confidence levels, redundancy of the candidate regions can be reduced, and the efficiency and the accuracy of remote sensing image detection are improved.
Fig. 3 illustrates a physical schematic diagram of an electronic device, as shown in fig. 3, where the electronic device may include: processor 310, communication interface (Communications Interface) 320, memory 330 and communication bus 340, wherein processor 310, communication interface 320 and memory 330 communicate with each other via communication bus 340. The processor 310 may invoke logic instructions in the memory 330 to perform an identification object candidate region screening method comprising: acquiring a candidate frame edge map of an image to be detected, and acquiring the number of complete contours included in a candidate frame according to the candidate frame edge map; obtaining a texture characteristic value of an image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object; and calculating texture complexity scores according to the number of complete contours and the entropy of the local binary pattern included in the candidate frame, taking the texture complexity scores as the confidence level of each candidate region, and screening the candidate regions of the identification object according to the confidence level.
In another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the recognition object candidate region screening method provided by the above methods, the method comprising: acquiring a candidate frame edge map of an image to be detected, and acquiring the number of complete contours included in a candidate frame according to the candidate frame edge map; obtaining a texture characteristic value of an image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object; and calculating texture complexity scores according to the number of complete contours and the entropy of the local binary pattern included in the candidate frame, taking the texture complexity scores as the confidence level of each candidate region, and screening the candidate regions of the identification object according to the confidence level.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A method for screening candidate regions of an identification object, comprising:
Acquiring a candidate frame edge map of an image to be detected, and acquiring the number of complete contours included in a candidate frame according to the candidate frame edge map;
Obtaining a texture characteristic value of the image to be detected, and calculating a local binary pattern entropy according to the texture characteristic value, wherein the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object;
And calculating texture complexity scores according to the number of complete contours included in the candidate frames and the local binary pattern entropy, taking the texture complexity scores as confidence measures of each candidate region, and screening out candidate regions of the identification object according to the confidence measures.
2. The method for screening candidate regions of an identification object according to claim 1, wherein the step of obtaining a candidate frame edge map of an image to be detected comprises:
And acquiring the edge intensity and the direction of each pixel point in the image to be detected based on a method of a structure forest, and determining an edge map of each candidate frame according to the edge intensity and the direction of each pixel point.
3. The method according to claim 1 or 2, wherein the obtaining the number of complete contours included in the candidate frame according to the candidate frame edge map includes:
Acquiring the sum of all edge map areas in each candidate frame, the width and the height of each candidate frame, and a weight function corresponding to the complete outline in each candidate frame;
and calculating the number of complete contours included in each candidate frame according to the sum of all edge map areas in each candidate frame, the width and the height of each candidate frame and the weight function.
4. A recognition object candidate region screening method as defined in claim 3 wherein the weight function is calculated based on the average directional affinity and length of each profile.
5. The method according to claim 1, wherein the obtaining the texture feature value of the image to be measured includes:
In a3×3 window, using the gray value of the pixel point at the center of the window as a threshold value, comparing the gray value of 8 adjacent pixels with the threshold value, if the gray value is larger than the threshold value, marking the position of the corresponding pixel point as 1, otherwise, setting the position as 0, generating 8-bit binary numbers, and taking the 8-bit binary numbers as texture characteristic values of the pixel point at the center of the window.
6. The method according to claim 1, wherein calculating the local binary pattern entropy from the texture feature value comprises:
Calculating the probability of each mode corresponding to the value of the local binary mode according to the value of the local binary mode and the width and the height of the candidate region;
and calculating the local binary mode entropy according to the probability of each mode.
7. The recognition object candidate region screening method of claim 1 or 6, wherein the computing a texture complexity score from the number of complete contours included in the candidate block and the local binary pattern entropy comprises:
Designing a gate function of the local binary pattern entropy, wherein the gate function comprises a plurality of extraction thresholds;
And constructing a texture complexity calculation formula according to the gate function and the number of the complete contours, extracting candidate frames meeting the proper area according to an extraction threshold, and calculating the texture complexity score of the candidate region.
8. An identification object candidate region screening device, characterized by comprising:
The first acquisition module is used for acquiring a candidate frame edge map of the image to be detected, and acquiring the number of complete contours included in the candidate frame according to the candidate frame edge map;
the second acquisition module is used for acquiring the texture characteristic value of the image to be detected, calculating a local binary pattern entropy according to the texture characteristic value, and the local binary pattern entropy is used for representing the difference between the identification object and the non-identification object;
and the screening module is used for calculating texture complexity scores according to the number of the complete contours included in the candidate frames and the local binary pattern entropy, using the texture complexity scores as confidence measures for measuring each candidate region, and screening the candidate regions of the identification objects according to the confidence measures.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the recognition object candidate region screening method of any one of claims 1 to 7 when the program is executed by the processor.
10. A non-transitory readable storage medium having stored thereon a computer program, which when executed by a processor implements the recognition object candidate region screening method of any one of claims 1 to 7.
CN202311775066.3A 2023-12-21 2023-12-21 Identification object candidate region screening method and device, electronic equipment and storage medium Pending CN117934909A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311775066.3A CN117934909A (en) 2023-12-21 2023-12-21 Identification object candidate region screening method and device, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311775066.3A CN117934909A (en) 2023-12-21 2023-12-21 Identification object candidate region screening method and device, electronic equipment and storage medium

Publications (1)

Publication Number Publication Date
CN117934909A true CN117934909A (en) 2024-04-26

Family

ID=90756484

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311775066.3A Pending CN117934909A (en) 2023-12-21 2023-12-21 Identification object candidate region screening method and device, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN117934909A (en)

Similar Documents

Publication Publication Date Title
Chaudhuri et al. Automatic building detection from high-resolution satellite images based on morphology and internal gray variance
Shi et al. Automatic road crack detection using random structured forests
WO2017190656A1 (en) Pedestrian re-recognition method and device
EP1233374B1 (en) Apparatus and method for extracting objects based on feature matching between segmented regions in images
CN113160192A (en) Visual sense-based snow pressing vehicle appearance defect detection method and device under complex background
CN104978567B (en) Vehicle checking method based on scene classification
CN106203539B (en) Method and device for identifying container number
CN109460735B (en) Document binarization processing method, system and device based on graph semi-supervised learning
WO2019197021A1 (en) Device and method for instance-level segmentation of an image
CN111275040A (en) Positioning method and device, electronic equipment and computer readable storage medium
CN110852327A (en) Image processing method, image processing device, electronic equipment and storage medium
Liang et al. A new wavelet-Laplacian method for arbitrarily-oriented character segmentation in video text lines
CN111274964B (en) Detection method for analyzing water surface pollutants based on visual saliency of unmanned aerial vehicle
Shi et al. Adaptive graph cut based binarization of video text images
Velliangira et al. A novel forgery detection in image frames of the videos using enhanced convolutional neural network in face images
Chen et al. Image segmentation based on mathematical morphological operator
CN117115117B (en) Pathological image recognition method based on small sample, electronic equipment and storage medium
CN111402185B (en) Image detection method and device
CN111104857A (en) Identity recognition method and system based on gait energy diagram
Ghandour et al. Building shadow detection based on multi-thresholding segmentation
Feng et al. Image shadow detection and removal based on region matching of intelligent computing
Wu et al. Vehicle detection in high-resolution images using superpixel segmentation and CNN iteration strategy
CN113920147B (en) Remote sensing image building extraction method and device based on deep learning
CN116030472A (en) Text coordinate determining method and device
CN115661660A (en) SAR image bridge rapid detection method and equipment

Legal Events

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