CN111859004A - Retrieval image acquisition method, device, equipment and readable storage medium - Google Patents

Retrieval image acquisition method, device, equipment and readable storage medium Download PDF

Info

Publication number
CN111859004A
CN111859004A CN202010743641.1A CN202010743641A CN111859004A CN 111859004 A CN111859004 A CN 111859004A CN 202010743641 A CN202010743641 A CN 202010743641A CN 111859004 A CN111859004 A CN 111859004A
Authority
CN
China
Prior art keywords
image
cluster center
index
similarity
query
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
CN202010743641.1A
Other languages
Chinese (zh)
Inventor
刘宏坤
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.)
Shuhang Technology Beijing Co ltd
Original Assignee
Shuhang Technology Beijing 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 Shuhang Technology Beijing Co ltd filed Critical Shuhang Technology Beijing Co ltd
Priority to CN202010743641.1A priority Critical patent/CN111859004A/en
Publication of CN111859004A publication Critical patent/CN111859004A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
    • G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/53—Querying
    • G06F16/538—Presentation of query results
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/55—Clustering; Classification

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Databases & Information Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Library & Information Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Processing Or Creating Images (AREA)

Abstract

The application provides a retrieval image obtaining method, which comprises the steps of determining the cluster center similarity of each index image and a query image, determining candidate index images according to the cluster center similarity, and further determining the retrieval image of the query image from the candidate index images based on the image similarity. It can be understood that the cluster center similarity is the number of the same similar cluster centers of the retrieval image and the query image, and the calculation process of the number is less computationally intensive than the similarity calculation of the feature vector, and the calculation process of the cluster centers and the similar cluster centers is also less computationally intensive than the similarity of the feature vector, so that the embodiment obtains the candidate index image with less computational effort. Compared with the prior art, on one hand, the retrieval range is reduced for all index images. On the other hand, the image similarity of each index image and each query image does not need to be calculated, so that the image retrieval efficiency is greatly improved.

Description

Retrieval image acquisition method, device, equipment and readable storage medium
Technical Field
The present application relates to the field of image retrieval technologies, and in particular, to a method, an apparatus, a device, and a readable storage medium for retrieving an image.
Background
In an image search engine, all index images containing the same landmarks as a given query image need to be retrieved from a specified image index library. In the prior art, based on the feature vector of the query image and the feature vector of each index image in the image index library, similarity analysis is performed on the query image and the feature vectors of all index images, and an index image containing the same landmark as the query image is retrieved according to the similarity analysis result.
Obviously, in the scene of the image index library with huge magnitude, the prior art needs to perform a massive image similarity analysis process, and the image retrieval speed is slow and the efficiency is low.
Disclosure of Invention
In view of the above, the present application provides a method, an apparatus, a device and a readable storage medium for obtaining a retrieval image, which are used to improve the speed and efficiency of image retrieval, and the following steps:
an acquisition method of a retrieval image, comprising:
clustering local feature vectors of the index images to obtain cluster center vectors;
acquiring a first similar cluster center and a second similar cluster center, wherein the first similar cluster center is the cluster center vector which is closest to each local feature vector of the index image, and the second similar cluster center is the cluster center vector which is closest to each local feature vector of the query image;
determining candidate index images of the query image based on cluster center similarity, wherein the cluster center similarity is the number of cluster center vectors in the first similar cluster center, which are the same as the second similar cluster center;
and determining a retrieval image of the query image from the candidate index images based on the image similarity.
Optionally, determining a candidate index image of the query image based on the cluster center similarity includes:
calculating the cluster center similarity of each index image and the query image;
sorting the index images according to the cluster core similarity to obtain a sorting result;
and taking N preset index images in the sequencing result as the candidate index images.
Optionally, determining a retrieval image of the query image from the candidate index images based on image similarity comprises:
calculating the image similarity of each candidate index image and the query image;
and determining the candidate index image with the image similarity with the query image larger than a preset first threshold value as a retrieval image of the query image.
Optionally, after the determining, based on the image similarity, a retrieval image of the query image from the candidate index images, the method further includes:
determining an optimal retrieval image from the retrieval images of the query image according to preset conditions, wherein the preset conditions comprise: a first condition and/or a second condition;
the first condition includes: the number of the matching feature point pairs included in the retrieval image and the query image is greater than a preset second threshold, and the second condition includes: matching feature point pairs included in the retrieval image and the query image satisfy linear transformation; any pair of matching feature point pairs included in the retrieval image and the query image includes: the search image feature extraction method comprises a first feature point and a second feature point, wherein the first feature point is any feature point in the search image, and the second feature point is a feature point matched with the first feature point in the query image.
Optionally, sorting the index images according to the cluster center similarity to obtain a sorting result, where the sorting result includes:
and sorting the index images with the cluster center similarity not less than a preset third threshold value according to the cluster center similarity to obtain a sorting result.
Optionally, the extracting process of the local feature vector includes:
inputting an image into a preset feature model, and using the output of the feature model as a local feature vector of the image, wherein the image comprises the query image and/or the index image.
Optionally, the method for acquiring the index image includes:
and inputting the images in the image index library into a preset classification model, and determining the images of which the classification results output by the classification model are landmark images as the index images.
An acquisition apparatus that retrieves an image, comprising:
the clustering unit is used for clustering the local characteristic vectors of the index images to obtain cluster center vectors;
a similar cluster center obtaining unit, configured to obtain a first similar cluster center and a second similar cluster center, where the first similar cluster center is the cluster center vector closest to each local feature vector of the index image, and the second similar cluster center is the cluster center vector closest to each local feature vector of the query image;
a candidate image determining unit, configured to determine a candidate index image of the query image based on the cluster center similarity, where the cluster center similarity is the number of cluster center vectors in the first similar cluster center that are the same as the second similar cluster center;
and the retrieval image determining unit is used for determining the retrieval image of the query image from the candidate index images based on the image similarity.
An acquisition apparatus that retrieves an image, comprising: a memory and a processor;
the memory is used for storing programs;
the processor is configured to execute the program to implement the steps of the method for acquiring a search image as described above.
A readable storage medium on which a computer program is stored, characterized in that the computer program, when being executed by a processor, carries out the steps of the acquisition method of a search image as described above.
According to the technical method, the method for acquiring the retrieval image determines the cluster similarity of each index image and the query image, determines the candidate index images according to the cluster similarity, and further determines the retrieval image from the candidate index images based on the image similarity. It can be understood that the cluster center similarity is the number of similar cluster centers between the search image and the query image, and the calculation process of the number consumes less calculation power compared with the calculation of the similarity of the feature vector, and the calculation process of the cluster centers and the similar cluster centers also saves calculation power compared with the similarity of the feature vector, so that the candidate index image is obtained with less calculation power in the embodiment. Compared with the prior art, on one hand, the retrieval range is reduced for all index images. On the other hand, the image similarity of each index image and each query image does not need to be calculated, so that the image retrieval efficiency is greatly improved. Particularly, under the image index library with the number of contained index images reaching the million orders, the image similarity calculation process is effectively reduced, the image retrieval speed is improved, and therefore the index images containing the same landmarks as the query images, namely the retrieval images of the query images, are rapidly obtained.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a schematic flowchart of a specific implementation method for acquiring a search image according to an embodiment of the present disclosure;
fig. 2 is a schematic flow chart of another specific implementation method for acquiring a search image according to an embodiment of the present application;
fig. 3 is a schematic flowchart of an acquisition method for retrieving an image according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of an apparatus for retrieving an image according to an embodiment of the present disclosure;
fig. 5 is a schematic structural diagram of an acquisition apparatus for retrieving an image according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
An application scenario of the retrieval image obtaining method provided by the embodiment of the application is that, according to a given query image, an index image containing the same landmark as the query image is retrieved from a preset image index library and is marked as a retrieval image. It is understood that the retrieval image refers to an index image that contains the same landmark as the query image. Taking the query image a as an example, and the landmark included in the query image a is a pyramid, the method can retrieve the index image including the pyramid from the preset image index library, and it can be understood that the number of the retrieved images may be one or more.
In order to improve the retrieval speed of an image, an embodiment of the present application provides an obtaining method of a retrieved image, and fig. 1 is a specific implementation method of the obtaining method of the retrieved image provided by the embodiment of the present application, which may specifically include the following steps:
s101, acquiring an index image.
When the non-landmark image is included in the preset image index library, a large amount of calculation power is wasted in acquiring the retrieval image for the non-landmark image, and the time for image retrieval is increased. Therefore, the embodiment performs filtering on non-landmark images with respect to a preset image index library, and obtains an index image.
Specifically, the images in the image index library are input to a classification model, and a classification result is output by the classification model, where the classification model is a pre-trained image classification network, and the classification result output by the image classification network includes a landmark image or a non-landmark image.
It can be appreciated that this step can improve the efficiency of image retrieval by reducing the order of magnitude of the index image by excluding non-landmark images.
When the preset image index library is the landmark image index library, that is, the images in the preset image index library are all landmark images, the filtering operation is not needed.
And S102, extracting a local feature vector of the query image.
Specifically, for any image, the local feature of the image refers to a feature of a feature point that appears stably in the image and has good distinguishability, and in general, the feature of a characteristic point can be represented by a local feature vector.
In this embodiment, a query image is input to a feature model, and a local feature vector of the query image is extracted by the feature model, where the feature model is a deep learning network trained in advance, and for the query image, the number of local feature vectors of the query image generated by the deep learning network may include a plurality of local feature vectors.
In this embodiment, the number of the generated local feature vectors of the query image is r, and is denoted as E1，E2，…，Er. Querying any local feature vector (E) of an image1、E2…, or Er) All dimensions of (A) are B. For example, the query image Q1 is input to a feature model that outputs 10 local feature vectors E of Q11、E2A10The dimension of any local feature vector is B1, and it should be noted that B1 may be any integer, for example, B1 is 100.
S103, extracting local characteristic vectors of the index images.
Specifically, the present embodiment inputs the index image to the feature model, and extracts the local feature vector of the index image from the feature model. For any index image, the number of local feature vectors of the index image generated by the feature model may include a plurality. It should be noted that the number of local feature vectors of any index image may be the same as or different from the number of local feature vectors of the query image, and the number of local feature vectors of different index images may be the same as or different from each other. In the embodiment of the present application, for convenience of description, the number of local feature vectors of any index image is k.
In this embodiment, let any index image be Si, where i ∈ [1, M ∈ [ ]]And M is the number of index images. The number of local feature vectors Wi of the index image Si generated by the feature model is k and is respectively recorded as Wi1，Wi2，…，Wik。
It should be noted that, in this embodiment, the dimensions of any local feature vector of any one of the aforementioned images (including the query image or any index image) are the same.
It should be further noted that, in the embodiment of the present application, the feature model for extracting the local feature vector of the query image and the feature model for extracting the local feature vector of the index image may be trained by the same deep learning network, and the training process of the deep learning network may refer to the prior art.
It can be understood that the local feature vector of any image (including the query image or any index image) extracted by the trained deep learning network in the step can characterize the local feature of the image, and any local feature has good distinguishability for different images.
And S104, clustering the local feature vectors of all the index images to obtain P cluster center vectors.
Specifically, the number of the index images is M, and the number of the generated local feature vectors of each index image is k, then the embodiment clusters the M × k local feature vectors to obtain P cluster center vectors, and numbers the P cluster center vectors, thereby obtaining any cluster center vector OjCluster center number of oj，j∈[1,P]. Wherein any cluster center vector OjDimension of (d) is B.
The specific implementation manner of the clustering method in this embodiment may refer to the prior art.
S105, determining the cluster center number of the cluster center vector (marked as the nearest cluster center vector of the local feature vector) with each local feature vector of the query image closest to the local feature vector, and the distance between each local feature vector and the nearest cluster center vector, and taking the cluster center number and the distance as the first feature parameter of the query image.
That is, the first feature parameters of the query image include: and querying the cluster center number of the nearest cluster center vector of each local feature vector of the image, and the distance between the local feature vector and the nearest cluster center vector.
For example, the local feature vector of the query image Q is E1，E2，…，ErWith E1For example, calculate E1The distance to each cluster center vector is compared to the magnitude of the distance. To obtain a compound of formula (II)1The nearest cluster center vector is O3And E is1And O3Is d1, then in this embodiment E will be1Is represented by (o)3D1), wherein o3Is the radial amount of the cluster center O3Cluster center number of (1).
It should be noted that, in this embodiment, the nearest cluster center vector of any local feature vector of the query image is recorded as a similar cluster center of the query image. It will be appreciated that the vector O of the cluster center3A similar cluster center for query image Q.
With reference to the above method, E1Are respectively replaced by E2、E3…、ErEach local feature vector of the query image may be represented as a cluster center number of a nearest cluster center vector of the local feature vector, and a distance of the local feature vector from the nearest cluster center vector.
S106, determining the cluster center number of the cluster center vector (marked as the nearest cluster center vector of the local feature vector) with each local feature vector of each index image closest to the local feature vector, and the distance between each local feature vector and the nearest cluster center vector, and taking the cluster center number and the distance as the first feature parameters of the index image.
That is, the first feature parameters of any index image include: the cluster center number of the nearest cluster center vector of each local feature vector of the index image, and the distance between the local feature vector and the nearest cluster center vector.
Taking index image S1 as an example, index image S1The partial feature vector is W11，W12，…，W1kW11For example, calculate W11The distance to each cluster center vector is compared to the magnitude of the distance. To obtain a product of formula W11The nearest cluster center vector is O2，W11And O2D2, then W1 in this embodiment1Is represented by (o)2D2), wherein o2Is the radial amount of the cluster center O2Cluster center number of (1).
It should be noted that, in this embodiment, the nearest cluster center vector of any local feature vector of the index image is recorded as a similar cluster center of the index image. It will be appreciated that the vector O of the cluster center2Is a similar cluster center of the index image S1.
With reference to the above method, W11Respectively replaced by W12、…、W1kEach local feature vector of the index image S1 can be expressed as a cluster center number of a cluster center vector that is closest to the local feature vector and a distance of the local feature vector from its closest cluster center vector.
It should be noted that, referring to the above method, in this step, each local feature vector of each index image can be represented as a cluster center number of a nearest cluster center vector of the local feature vector and a distance between the local feature vector and the nearest cluster center vector.
S107, determining the cluster center similarity of each index image and the query image, and taking the N index images with higher cluster center similarity with the query image as candidate index images.
In this embodiment, for any index image Si, the similarity of the cluster centers of Si and the query image Q is denoted as IiThen, IiIs the number of cluster center vectors in the similar cluster center of Si that are identical to the similar cluster center of Q. Wherein, the similar cluster center of Si refers to: the nearest cluster center vector for each local feature vector of Si. The similar cluster centers of Q refer to: the nearest cluster center vector for each local feature vector of Q. It is understood that the number of similar cluster centers of any image may include a plurality.
The specific implementation manner of determining the cluster center similarity between the index image and the query image and determining the candidate index image based on the cluster center similarity may include multiple manners. Specifically, the method comprises the following steps A1-A3:
a1, representing the similar cluster center (i.e. the first similar cluster center) of each index picture and the similar cluster center (i.e. the first similar cluster center) of the query picture in the form of text or an array, wherein in practical applications, the representation form of the similar cluster centers can be selected based on the number of the index pictures, for example, when the number of the index pictures reaches the order of millions, to improve the retrieval speed, the similar cluster center of each index picture is represented in the form of text, and in the text, the cluster center numbers of the similar cluster centers of the index pictures are separated by commas, thereby obtaining the text representation form of each index picture.
A2, determining the number of cluster center vectors in the similar cluster center of each index image, which is the same as the similar cluster center of the query image, and determining the number as the cluster center similarity of the index image and the query image.
It should be noted that, a specific implementation process of the search method for the inverted index may refer to the prior art, and details are not described in this embodiment.
And A3, sorting the index images from large to small according to the similarity of the index images and the cluster centers of the query images, and selecting the first N index images as candidate index images. Wherein, N can be set according to actual needs.
It should be noted that, the specific implementation manner for determining the candidate index image based on the cluster center similarity may include multiple manners, for example, an index image whose cluster center similarity with the query image exceeds the threshold Φ may also be selected as the candidate index image.
It can be understood that the cluster center similarity between any index image and the query image can represent the similarity between the index image and the query image, that is, when the cluster center similarity between the index image and the query image is high, the index image and the query image are considered to be highly similar. Therefore, the N candidate index images selected in this step are N index images with higher cluster center similarity to the query image, that is, the index images with low similarity to the query image are excluded in this step. Based on this, the present embodiment further selects an index image that is the same as the landmark information included in the query image from among the N candidate index images.
S108, calculating the image similarity of each candidate index image and the query image, and determining the candidate index image with the image similarity larger than a preset first threshold value as a target index image.
In this embodiment, the image similarity between each index image and the query image is ASMK (Aggregate select Match kernel) similarity between the index image and the query image, and the ASMK similarity can more accurately represent the similarity between the two images. Generally, the higher the ASMK similarity of the index image and the query image, the higher the similarity of the index image and the query image, and the higher the probability of containing the same landmark. With ASMK similarity I of query image Q and index image S1ASMKFor example, IASMKCalculated according to the first characteristic parameters of the query image Q and the first characteristic parameters of the index image S1.
It should be noted that, for a specific implementation manner of calculating the ASMK similarity between the candidate index image and the query image, reference may be made to the prior art.
And S109, taking the target index image as a retrieval image of the query image.
It is understood that the target index image may include a plurality of target index images, and each target index image has an ASMK similarity with the query image that exceeds a preset first threshold, so that the target index image may be directly used as the search image of the query image.
It can be seen from the foregoing technical methods that, in the retrieval image obtaining method provided in the embodiment of the present application, the cluster center similarity between each index image and the query image is determined, the candidate index images are determined based on the cluster center similarity, and the retrieval images are further determined based on the image similarity in the candidate index images. It can be understood that the cluster center similarity is the number of similar cluster centers between the search image and the query image, and the calculation process of the number consumes less calculation power compared with the calculation of the similarity of the feature vector, and the calculation process of the cluster centers and the similar cluster centers also saves calculation power compared with the similarity of the feature vector, so that the candidate index image is obtained with less calculation power in the embodiment. Compared with the prior art, on one hand, the retrieval range is reduced for all index images. On the other hand, the image similarity of each index image and each query image does not need to be calculated, so that the image retrieval efficiency is greatly improved. Particularly, under the image index library with the number of contained index images reaching the million orders, the image similarity calculation process is effectively reduced, the image retrieval speed is improved, and therefore the index images containing the same landmarks as the query images, namely the retrieval images of the query images, are rapidly obtained.
Further, compared with a conventional method using a global feature vector in the prior art, the method for acquiring a search image provided by the embodiment of the present application performs calculation of cluster center similarity and ASMK similarity based on local features of an image, and has high accuracy and recall rate. In addition, the embodiment of the application firstly filters the non-landmark images to exclude the non-landmark images, further reduces the retrieval range and improves the speed of image retrieval.
It should be noted that, optionally, the target index image may be further processed to obtain a retrieval image with higher accuracy. Fig. 2 is another retrieval image obtaining method provided in the embodiment of the present application, which may specifically include:
s201, acquiring an index image.
S202, extracting local feature vectors of the query image.
And S203, extracting the local feature vector of the index image.
And S204, clustering the local feature vectors of all the index images to obtain P cluster center vectors.
S205, determining a cluster center number of a cluster center vector (marked as the nearest cluster center vector of the local feature vector) of each local feature vector of the query image, and a distance between each local feature vector and the nearest cluster center vector, and taking the cluster center number and the distance as a first feature parameter of the query image.
S206, determining the cluster center number of the cluster center vector (marked as the nearest cluster center vector of the local feature vector) with each local feature vector of each index image closest to the local feature vector, and the distance between each local feature vector and the nearest cluster center vector, and taking the cluster center number and the distance as the first feature parameters of the index image.
S207, determining the cluster center similarity of each index image and the query image, and taking the N index images with higher cluster center similarity with the query image as candidate index images.
S208, calculating the image similarity between each candidate index image and the query image, and determining the candidate index image with the image similarity larger than a preset first threshold value as the target index image.
It should be noted that specific implementation manners of S201 to S208 may refer to S101 to S108, which are not described in detail in this embodiment.
S209, calculating the number of the matching characteristic point pairs of each target index image and the query image, and searching the target index images of which the number of the matching characteristic point pairs with the query image is greater than a preset second threshold value.
The matching feature point pairs of the target index image and the query image comprise: the target indexes a local feature vector of the image and a local feature vector of the query image that matches the local feature vector. It should be noted that, the method for determining whether any local feature vector of the target index image matches any local feature vector of the query image may be to determine that the local feature vectors with ASMK similarity greater than a preset matching threshold are matched, and may specifically refer to the prior art.
In this embodiment, after the number of the matching feature point pairs of each target index image and the query image is determined, the number of the matching feature point pairs is compared with a preset second threshold, so as to obtain a target index image in which the number of the matching feature point pairs of the query image is greater than the preset second threshold.
And S210, for each target index image obtained in the S209, randomly extracting K matching feature point pairs of the target index image and the query image, judging whether the K matching feature point pairs meet linear transformation, and if so, taking the target index image as a retrieval image of the query image.
In this embodiment, the value of K may be preset according to an actual situation. It should be noted that, for a specific implementation of determining whether K matching feature point pairs satisfy linear transformation, reference may be made to the prior art.
It should be noted that the retrieval image obtaining method provided by the embodiment of the present application may be applied to an image retrieval system for obtaining a retrieval image, where the image retrieval system may be provided with a plurality of cloud ends, and each cloud end may perform cloud storage on data generated in the retrieval image obtaining process. For example, a first cloud is provided for storing the id of all images (query image and each index image) and the similar cluster center of the image. And setting a second cloud end for storing the image ID of the candidate index image. And setting a third cloud end for storing the target index image.
In summary, in this embodiment, through S209 to S210, geometric verification is performed on each target index image and the query image, and the target index image whose geometric verification result satisfies the preset verification condition is determined as the search image of the query image. The preset checking condition comprises a first condition and a second condition, wherein the first condition is as follows: the number of matching feature points of the target index image and the query image is greater than a preset second threshold, and the second condition is as follows: and matching characteristic point pairs of the target index image and the query image satisfy linear transformation. Therefore, the target index image is further retrieved in the embodiment of the present application, since the larger the number of the matching feature point pairs is, the higher the similarity between the target index image and the query image is, and the higher the linearity of the matching feature point pairs is, the higher the similarity between the target index image and the query image is, the accuracy of image retrieval can be further improved through the above geometric verification process.
Furthermore, the retrieval image acquisition method provided by the embodiment of the application carries out cloud storage on the data by establishing one or more cloud ends, so that the situation that a large amount of computing power and time are consumed by repeatedly computing the same data can be avoided, and the retrieval efficiency is improved.
Fig. 1 and fig. 2 are specific embodiments of acquiring a search image according to an embodiment of the present application, and fig. 3 is a flowchart of a method for acquiring a search image according to an embodiment of the present application, and as shown in fig. 3, the present embodiment summarizes the flows in the above embodiments as follows, i.e., S301 to S304.
S301, clustering the local feature vectors of the index images to obtain cluster center vectors.
The index image may be an image in a preset image index library, or may be a filtered landmark image, and the specific method for filtering a non-landmark image may refer to S101.
The method for obtaining the local feature vector of any index image comprises the steps of inputting the index image into a preset feature model, and taking the output of the feature model as the local feature vector of the index image. A specific method of indexing the local feature vector of the image may refer to S102 described above.
It is understood that the local feature vector of any index image in this step may include a plurality of local feature vectors, where any local feature vector may characterize the attribute of the local feature point of the index image. In the step, all local feature vectors of all index images are clustered, so that a plurality of cluster center vectors can be obtained. It should be noted that, the process of acquiring the cluster center vector may refer to S104 described above.
S302, obtaining a first similar cluster center and a second similar cluster center.
The first similar cluster center is a cluster center vector closest to each local feature vector of the index image, and the second similar cluster center is a cluster center vector closest to each local feature vector of the query image.
Specifically, for any index image, the distance between each local feature vector and each cluster center vector of the index image is calculated, for each local feature vector, the cluster center vector closest to the local feature vector is selected, and the cluster center vector is used as the similar cluster center of the index image, namely the first similar cluster center. It will be appreciated that the first similar cluster of each index image comprises a plurality of cluster center vectors.
And aiming at the query image, calculating the distance between each local feature vector of the query image and each cluster center vector, selecting the cluster center vector with the closest distance for each local feature vector, and taking the cluster center vector as a similar cluster center of the query image, namely a second similar cluster center. It is understood that the second similar cluster includes a plurality of cluster center vectors.
It should be noted that, the method for obtaining the local feature vector of the query image may refer to S102 described above.
S303, determining candidate index images of the query image based on the cluster center similarity.
And the cluster center similarity is the number of the cluster center vectors in the first similar cluster center, which are the same as the second similar cluster center. It can be understood that each cluster center vector is obtained by clustering local feature vectors of the index image, and when any two local feature vectors are closest to the same cluster center vector, the attributes of the feature points represented by the two local feature vectors are close. Therefore, the cluster center similarity may represent how similar the query image is to the index image. Therefore, the candidate index image determined in the step is the index image with higher similarity degree with the query image based on the cluster center similarity.
The specific method for determining the candidate index image may include various methods, and an optional method includes steps B1 to B3.
And B1, calculating the cluster center similarity of each index image and the query image.
B2, sorting the index images according to the cluster center similarity to obtain a sorting result.
And B3, taking N index images preset in the sorting result as candidate index images. The value of N is preset according to an actual situation, and the N index images may be N index images ranked at the top in the sorting result.
In another alternative method, after B1, the index images whose cluster center similarity with the query image is not less than the preset third threshold may be sorted according to the cluster center similarity to obtain a sorting result, and the N index images preset in the sorting result may be further used as candidate index images.
It should be noted that, for specific implementation manners of the steps B1 to B3, reference may be made to S107 in the foregoing embodiment.
S304, determining a retrieval image of the query image from the candidate index images based on the image similarity.
The image similarity between any index image and the query image may be ASMK similarity between the index image and the query image, and it should be noted that ASMK similarity is a metric for more accurately representing the similarity between two images. Specific calculation methods can be found in the prior art.
The image similarity of each candidate index image and the query image is calculated, and the candidate index image with the image similarity larger than a preset first threshold value is determined as the retrieval image of the query image.
Further, the embodiment may also determine an optimal search image from the search images of the query image according to a preset condition. Wherein the preset conditions include: a first condition and/or a second condition.
Specifically, the first condition includes: the number of the matched feature point pairs included in the retrieval image and the query image is larger than a preset second threshold value. The second condition includes: the retrieval image and the matching feature point pairs included in the query image satisfy linear transformation.
It should be noted that any pair of matching feature point pairs included in the search image and the query image includes: the image retrieval method comprises a first characteristic point and a second characteristic point, wherein the first characteristic point is any characteristic point in a retrieval image, and the second characteristic point is a characteristic point which is matched with the first characteristic point in a query image. Wherein any feature point can be represented by a local feature vector.
It should be noted that, a specific process of determining whether any search image satisfies the first condition may refer to S209, and a specific process of determining whether any search image satisfies the second condition may refer to S210, which is not described in detail in this embodiment.
In summary, in another optional case, the optimal retrieval image selected by the method simultaneously satisfies: the ASMK similarity with the query image is larger than a preset first threshold, the number of the matched characteristic point pairs with the query image is not smaller than a preset second threshold, and the matched characteristic point pairs with the query image meet linear transformation.
According to the technical scheme, the method for acquiring the retrieval images determines the cluster similarity of each index image and the query image, determines the candidate index images according to the cluster similarity, and further determines the retrieval images in the candidate index images based on the image similarity. It can be understood that the cluster center similarity is the number of similar cluster centers between the search image and the query image, and the calculation process of the number is less computationally intensive, so that the embodiment uses less computational effort to obtain the candidate index image. Compared with the prior art, on one hand, the retrieval range is reduced for all index images. On the other hand, the image similarity of each index image and each query image does not need to be calculated, so that the image retrieval efficiency is greatly improved. Particularly, under the image index library with the number of contained index images reaching the million orders, the image similarity calculation process is effectively reduced, the image retrieval speed is improved, and therefore the index images containing the same landmarks as the query images, namely the retrieval images of the query images, are rapidly obtained.
The following describes the apparatus for acquiring a search image provided in the embodiments of the present application, and the apparatus for acquiring a search image described below and the method for acquiring a search image described above may be referred to in correspondence with each other.
Referring to fig. 4, a schematic structural diagram of an apparatus for acquiring a search image according to an embodiment of the present application is shown, and as shown in fig. 4, the apparatus may include:
a clustering unit 401, configured to cluster the local feature vectors of the index images to obtain cluster center vectors;
a similar cluster center obtaining unit 402, configured to obtain a first similar cluster center and a second similar cluster center, where the first similar cluster center is the cluster center vector closest to each local feature vector of the index image, and the second similar cluster center is the cluster center vector closest to each local feature vector of the query image;
a candidate image determining unit 403, configured to determine a candidate index image of the query image based on the cluster center similarity, where the cluster center similarity is the number of cluster center vectors in the first similar cluster center that are the same as the second similar cluster center;
a retrieval image determining unit 404, configured to determine a retrieval image of the query image from the candidate index images based on the image similarity.
Optionally, the candidate image determining unit is configured to determine a candidate index image of the query image based on the cluster center similarity, and includes: the candidate image determination unit is specifically configured to:
calculating the cluster center similarity of each index image and the query image;
sorting the index images according to the cluster core similarity to obtain a sorting result;
and taking N preset index images in the sequencing result as the candidate index images.
Optionally, the search image determining unit is configured to determine a search image of the query image from the candidate index images based on image similarity, and includes: the retrieval image determining unit is specifically configured to:
calculating the image similarity of each candidate index image and the query image;
and determining the candidate index image with the image similarity with the query image larger than a preset first threshold value as a retrieval image of the query image.
Optionally, the apparatus further includes an optimal image determining unit, configured to determine an optimal search image from the search images of the query image according to preset conditions after determining the search image of the query image from the candidate index images based on the image similarity, where the preset conditions include: a first condition and/or a second condition;
the first condition includes: the number of the matching feature point pairs included in the retrieval image and the query image is greater than a preset second threshold, and the second condition includes: matching feature point pairs included in the retrieval image and the query image satisfy linear transformation; any pair of matching feature point pairs included in the retrieval image and the query image includes: the search image feature extraction method comprises a first feature point and a second feature point, wherein the first feature point is any feature point in the search image, and the second feature point is a feature point matched with the first feature point in the query image.
Optionally, the candidate image determining unit is configured to rank the index images according to the cluster similarity, so as to obtain a ranking result, and includes: the candidate image determination unit is specifically configured to:
and sorting the index images with the cluster center similarity not less than a preset third threshold value according to the cluster center similarity to obtain a sorting result.
Optionally, the apparatus further includes a local feature extraction unit, configured to input an image to a preset feature model, and output the feature model as a local feature vector of the image, where the image includes the query image and/or the index image.
Optionally, the apparatus further includes an index image obtaining unit, configured to input an image in an image index library to a preset classification model, and determine an image of which a classification result output by the classification model is a landmark image as the index image.
An embodiment of the present application further provides a device for acquiring a search image, please refer to fig. 5, which shows a schematic structural diagram of the device for acquiring a search image, and the device may include: at least one processor 501, at least one communication interface 502, at least one memory 503, and at least one communication bus 504;
in the embodiment of the present application, the number of the processor 501, the communication interface 502, the memory 503 and the communication bus 504 is at least one, and the processor 501, the communication interface 502 and the memory 503 complete the communication with each other through the communication bus 504;
the processor 501 may be a central processing unit CPU, or an application specific Integrated circuit asic, or one or more Integrated circuits configured to implement embodiments of the present invention, or the like;
the memory 503 may include a high-speed RAM memory, and may further include a non-volatile memory (non-volatile memory) or the like, such as at least one disk memory;
the memory stores a program, and the processor can execute the program stored in the memory to realize the retrieval image acquisition method provided by the embodiment of the application, and the retrieval image acquisition method comprises the following steps:
clustering local feature vectors of the index images to obtain cluster center vectors;
acquiring a first similar cluster center and a second similar cluster center, wherein the first similar cluster center is the cluster center vector which is closest to each local feature vector of the index image, and the second similar cluster center is the cluster center vector which is closest to each local feature vector of the query image;
determining candidate index images of the query image based on cluster center similarity, wherein the cluster center similarity is the number of cluster center vectors in the first similar cluster center, which are the same as the second similar cluster center;
and determining a retrieval image of the query image from the candidate index images based on the image similarity.
Optionally, determining a candidate index image of the query image based on the cluster center similarity includes:
calculating the cluster center similarity of each index image and the query image;
sorting the index images according to the cluster core similarity to obtain a sorting result;
and taking N preset index images in the sequencing result as the candidate index images.
Optionally, determining a retrieval image of the query image from the candidate index images based on image similarity comprises:
calculating the image similarity of each candidate index image and the query image;
and determining the candidate index image with the image similarity with the query image larger than a preset first threshold value as a retrieval image of the query image.
Optionally, after the determining, based on the image similarity, a retrieval image of the query image from the candidate index images, the method further includes:
determining an optimal retrieval image from the retrieval images of the query image according to preset conditions, wherein the preset conditions comprise: a first condition and/or a second condition;
the first condition includes: the number of the matching feature point pairs included in the retrieval image and the query image is greater than a preset second threshold, and the second condition includes: matching feature point pairs included in the retrieval image and the query image satisfy linear transformation; any pair of matching feature point pairs included in the retrieval image and the query image includes: the search image feature extraction method comprises a first feature point and a second feature point, wherein the first feature point is any feature point in the search image, and the second feature point is a feature point matched with the first feature point in the query image.
Optionally, sorting the index images according to the cluster center similarity to obtain a sorting result, where the sorting result includes:
and sorting the index images with the cluster center similarity not less than a preset third threshold value according to the cluster center similarity to obtain a sorting result.
Optionally, the extracting process of the local feature vector includes:
inputting an image into a preset feature model, and using the output of the feature model as a local feature vector of the image, wherein the image comprises the query image and/or the index image.
Optionally, the method for acquiring the index image includes:
and inputting the images in the image index library into a preset classification model, and determining the images of which the classification results output by the classification model are landmark images as the index images.
An embodiment of the present application further provides a readable storage medium, where the readable storage medium may store a computer program suitable for being executed by a processor, and when the computer program is executed by the processor, the method for acquiring a search image according to the embodiment of the present application is implemented as follows:
clustering local feature vectors of the index images to obtain cluster center vectors;
acquiring a first similar cluster center and a second similar cluster center, wherein the first similar cluster center is the cluster center vector which is closest to each local feature vector of the index image, and the second similar cluster center is the cluster center vector which is closest to each local feature vector of the query image;
determining candidate index images of the query image based on cluster center similarity, wherein the cluster center similarity is the number of cluster center vectors in the first similar cluster center, which are the same as the second similar cluster center;
and determining a retrieval image of the query image from the candidate index images based on the image similarity.
Optionally, determining a candidate index image of the query image based on the cluster center similarity includes:
calculating the cluster center similarity of each index image and the query image;
sorting the index images according to the cluster core similarity to obtain a sorting result;
and taking N preset index images in the sequencing result as the candidate index images.
Optionally, determining a retrieval image of the query image from the candidate index images based on image similarity comprises:
calculating the image similarity of each candidate index image and the query image;
and determining the candidate index image with the image similarity with the query image larger than a preset first threshold value as a retrieval image of the query image.
Optionally, after the determining, based on the image similarity, a retrieval image of the query image from the candidate index images, the method further includes:
determining an optimal retrieval image from the retrieval images of the query image according to preset conditions, wherein the preset conditions comprise: a first condition and/or a second condition;
the first condition includes: the number of the matching feature point pairs included in the retrieval image and the query image is greater than a preset second threshold, and the second condition includes: matching feature point pairs included in the retrieval image and the query image satisfy linear transformation; any pair of matching feature point pairs included in the retrieval image and the query image includes: the search image feature extraction method comprises a first feature point and a second feature point, wherein the first feature point is any feature point in the search image, and the second feature point is a feature point matched with the first feature point in the query image.
Optionally, sorting the index images according to the cluster center similarity to obtain a sorting result, where the sorting result includes:
and sorting the index images with the cluster center similarity not less than a preset third threshold value according to the cluster center similarity to obtain a sorting result.
Optionally, the extracting process of the local feature vector includes:
inputting an image into a preset feature model, and using the output of the feature model as a local feature vector of the image, wherein the image comprises the query image and/or the index image.
Optionally, the method for acquiring the index image includes:
and inputting the images in the image index library into a preset classification model, and determining the images of which the classification results output by the classification model are landmark images as the index images.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other.
The above description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (10)

1. An acquisition method of a search image, comprising:
clustering local feature vectors of the index images to obtain cluster center vectors;
acquiring a first similar cluster center and a second similar cluster center, wherein the first similar cluster center is the cluster center vector which is closest to each local feature vector of the index image, and the second similar cluster center is the cluster center vector which is closest to each local feature vector of the query image;
determining candidate index images of the query image based on cluster center similarity, wherein the cluster center similarity is the number of cluster center vectors in the first similar cluster center, which are the same as the second similar cluster center;
and determining a retrieval image of the query image from the candidate index images based on the image similarity.
2. The method for obtaining a search image according to claim 1, wherein the determining the candidate index image of the query image based on the cluster center similarity comprises:
calculating the cluster center similarity of each index image and the query image;
sorting the index images according to the cluster core similarity to obtain a sorting result;
and taking N preset index images in the sequencing result as the candidate index images.
3. The method according to claim 1, wherein the determining the search image of the query image from the candidate index images based on the image similarity comprises:
calculating the image similarity of each candidate index image and the query image;
and determining the candidate index image with the image similarity with the query image larger than a preset first threshold value as a retrieval image of the query image.
4. The method according to claim 1 or 3, further comprising, after determining the search image of the query image from the candidate index images based on the image similarity, the steps of:
determining an optimal retrieval image from the retrieval images of the query image according to preset conditions, wherein the preset conditions comprise: a first condition and/or a second condition;
the first condition includes: the number of the matching feature point pairs included in the retrieval image and the query image is greater than a preset second threshold, and the second condition includes: matching feature point pairs included in the retrieval image and the query image satisfy linear transformation; any pair of matching feature point pairs included in the retrieval image and the query image includes: the search image feature extraction method comprises a first feature point and a second feature point, wherein the first feature point is any feature point in the search image, and the second feature point is a feature point matched with the first feature point in the query image.
5. The method for obtaining the search image according to claim 2, wherein the sorting the index images according to the cluster similarity to obtain a sorting result includes:
and sorting the index images with the cluster center similarity not less than a preset third threshold value according to the cluster center similarity to obtain a sorting result.
6. The method for acquiring retrieval image according to claim 1, wherein the process of extracting the local feature vector includes:
inputting an image into a preset feature model, and using the output of the feature model as a local feature vector of the image, wherein the image comprises the query image and/or the index image.
7. The method for acquiring a search image according to claim 1, wherein the method for acquiring an index image includes:
and inputting the images in the image index library into a preset classification model, and determining the images of which the classification results output by the classification model are landmark images as the index images.
8. An acquisition apparatus for retrieving an image, comprising:
the clustering unit is used for clustering the local characteristic vectors of the index images to obtain cluster center vectors;
a similar cluster center obtaining unit, configured to obtain a first similar cluster center and a second similar cluster center, where the first similar cluster center is the cluster center vector closest to each local feature vector of the index image, and the second similar cluster center is the cluster center vector closest to each local feature vector of the query image;
a candidate image determining unit, configured to determine a candidate index image of the query image based on the cluster center similarity, where the cluster center similarity is the number of cluster center vectors in the first similar cluster center that are the same as the second similar cluster center;
and the retrieval image determining unit is used for determining the retrieval image of the query image from the candidate index images based on the image similarity.
9. An acquisition apparatus that retrieves an image, comprising: a memory and a processor;
the memory is used for storing programs;
the processor is configured to execute the program to implement the steps of the retrieval image acquisition method according to any one of claims 1 to 7.
10. A readable storage medium on which a computer program is stored, the computer program, when executed by a processor, implementing the steps of the retrieval image acquisition method according to any one of claims 1 to 7.
CN202010743641.1A 2020-07-29 2020-07-29 Retrieval image acquisition method, device, equipment and readable storage medium Pending CN111859004A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010743641.1A CN111859004A (en) 2020-07-29 2020-07-29 Retrieval image acquisition method, device, equipment and readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010743641.1A CN111859004A (en) 2020-07-29 2020-07-29 Retrieval image acquisition method, device, equipment and readable storage medium

Publications (1)

Publication Number Publication Date
CN111859004A true CN111859004A (en) 2020-10-30

Family

ID=72945935

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010743641.1A Pending CN111859004A (en) 2020-07-29 2020-07-29 Retrieval image acquisition method, device, equipment and readable storage medium

Country Status (1)

Country Link
CN (1) CN111859004A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112269894A (en) * 2020-12-03 2021-01-26 北京沃东天骏信息技术有限公司 Article pool generation method, image search method, device, electronic equipment and medium
CN112948613A (en) * 2020-12-01 2021-06-11 上海云从企业发展有限公司 Image incremental clustering method, system, medium and device
CN113159211A (en) * 2021-04-30 2021-07-23 杭州好安供应链管理有限公司 Method, computing device and computer storage medium for similar image retrieval
CN113254687A (en) * 2021-06-28 2021-08-13 腾讯科技（深圳）有限公司 Image retrieval and image quantification model training method, device and storage medium
WO2023108995A1 (en) * 2021-12-15 2023-06-22 平安科技（深圳）有限公司 Vector similarity calculation method and apparatus, device and storage medium

Citations (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6285995B1 (en) * 1998-06-22 2001-09-04 U.S. Philips Corporation Image retrieval system using a query image
JP2010102629A (en) * 2008-10-27 2010-05-06 Kddi Corp Content retrieval system and content retrieval program
JP2010286861A (en) * 2009-06-09 2010-12-24 Yahoo Japan Corp Image retrieval device
CN103617217A (en) * 2013-11-20 2014-03-05 中国科学院信息工程研究所 Hierarchical index based image retrieval method and system
US20140105509A1 (en) * 2012-10-15 2014-04-17 Canon Kabushiki Kaisha Systems and methods for comparing images
CN103970769A (en) * 2013-01-29 2014-08-06 华为技术有限公司 Image retrieval method and image retrieval device
CN106777090A (en) * 2016-12-14 2017-05-31 大连交通大学 The medical science big data search method of the Skyline that view-based access control model vocabulary is matched with multiple features
CN106886553A (en) * 2016-12-27 2017-06-23 浙江宇视科技有限公司 A kind of image search method and server
CN107341151A (en) * 2016-04-29 2017-11-10 成都理想境界科技有限公司 Image retrieval data library generating method, the method and device of augmented reality
CN108334644A (en) * 2018-03-30 2018-07-27 百度在线网络技术（北京）有限公司 Image-recognizing method and device
CN109241327A (en) * 2017-07-03 2019-01-18 北大方正集团有限公司 Image search method and device
CN110110119A (en) * 2018-01-31 2019-08-09 株式会社理光 The method, apparatus and computer readable storage medium of image retrieval
CN110275970A (en) * 2019-06-21 2019-09-24 北京达佳互联信息技术有限公司 Method, apparatus, server and the storage medium of image retrieval
CN110458224A (en) * 2019-08-06 2019-11-15 北京字节跳动网络技术有限公司 Image processing method, device, electronic equipment and computer-readable medium
CN110674334A (en) * 2019-09-16 2020-01-10 南京信息工程大学 Near-repetitive image retrieval method based on consistency region deep learning features
CN110942046A (en) * 2019-12-05 2020-03-31 腾讯云计算（北京）有限责任公司 Image retrieval method, device, equipment and storage medium
CN110956195A (en) * 2019-10-11 2020-04-03 平安科技（深圳）有限公司 Image matching method and device, computer equipment and storage medium

Patent Citations (17)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6285995B1 (en) * 1998-06-22 2001-09-04 U.S. Philips Corporation Image retrieval system using a query image
JP2010102629A (en) * 2008-10-27 2010-05-06 Kddi Corp Content retrieval system and content retrieval program
JP2010286861A (en) * 2009-06-09 2010-12-24 Yahoo Japan Corp Image retrieval device
US20140105509A1 (en) * 2012-10-15 2014-04-17 Canon Kabushiki Kaisha Systems and methods for comparing images
CN103970769A (en) * 2013-01-29 2014-08-06 华为技术有限公司 Image retrieval method and image retrieval device
CN103617217A (en) * 2013-11-20 2014-03-05 中国科学院信息工程研究所 Hierarchical index based image retrieval method and system
CN107341151A (en) * 2016-04-29 2017-11-10 成都理想境界科技有限公司 Image retrieval data library generating method, the method and device of augmented reality
CN106777090A (en) * 2016-12-14 2017-05-31 大连交通大学 The medical science big data search method of the Skyline that view-based access control model vocabulary is matched with multiple features
CN106886553A (en) * 2016-12-27 2017-06-23 浙江宇视科技有限公司 A kind of image search method and server
CN109241327A (en) * 2017-07-03 2019-01-18 北大方正集团有限公司 Image search method and device
CN110110119A (en) * 2018-01-31 2019-08-09 株式会社理光 The method, apparatus and computer readable storage medium of image retrieval
CN108334644A (en) * 2018-03-30 2018-07-27 百度在线网络技术（北京）有限公司 Image-recognizing method and device
CN110275970A (en) * 2019-06-21 2019-09-24 北京达佳互联信息技术有限公司 Method, apparatus, server and the storage medium of image retrieval
CN110458224A (en) * 2019-08-06 2019-11-15 北京字节跳动网络技术有限公司 Image processing method, device, electronic equipment and computer-readable medium
CN110674334A (en) * 2019-09-16 2020-01-10 南京信息工程大学 Near-repetitive image retrieval method based on consistency region deep learning features
CN110956195A (en) * 2019-10-11 2020-04-03 平安科技（深圳）有限公司 Image matching method and device, computer equipment and storage medium
CN110942046A (en) * 2019-12-05 2020-03-31 腾讯云计算（北京）有限责任公司 Image retrieval method, device, equipment and storage medium

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
刘丹丹: ""基于加密词袋模型的密文图像检索研究"", no. 01 *

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112948613A (en) * 2020-12-01 2021-06-11 上海云从企业发展有限公司 Image incremental clustering method, system, medium and device
CN112948613B (en) * 2020-12-01 2023-09-15 上海云从企业发展有限公司 Image incremental clustering method, system, medium and device
CN112269894A (en) * 2020-12-03 2021-01-26 北京沃东天骏信息技术有限公司 Article pool generation method, image search method, device, electronic equipment and medium
CN112269894B (en) * 2020-12-03 2021-07-13 北京沃东天骏信息技术有限公司 Article pool generation method, image search method, device, electronic equipment and medium
CN113159211A (en) * 2021-04-30 2021-07-23 杭州好安供应链管理有限公司 Method, computing device and computer storage medium for similar image retrieval
CN113254687A (en) * 2021-06-28 2021-08-13 腾讯科技（深圳）有限公司 Image retrieval and image quantification model training method, device and storage medium
CN113254687B (en) * 2021-06-28 2021-09-17 腾讯科技（深圳）有限公司 Image retrieval and image quantification model training method, device and storage medium
WO2023108995A1 (en) * 2021-12-15 2023-06-22 平安科技（深圳）有限公司 Vector similarity calculation method and apparatus, device and storage medium

Similar Documents

Publication Publication Date Title
CN111859004A (en) Retrieval image acquisition method, device, equipment and readable storage medium
US11048966B2 (en) Method and device for comparing similarities of high dimensional features of images
CN110162695B (en) Information pushing method and equipment
EP2284791B1 (en) Method of creating three-dimensional object identifying image database, processing apparatus and processing program
CN110188223B (en) Image processing method and device and computer equipment
JP5121917B2 (en) Image search apparatus, image search method and program
US10482146B2 (en) Systems and methods for automatic customization of content filtering
JP5094830B2 (en) Image search apparatus, image search method and program
CN108629345B (en) High-dimensional image feature matching method and device
CN112732883A (en) Fuzzy matching method and device based on knowledge graph and computer equipment
CN110688565B (en) Next item recommendation method based on multidimensional Hox process and attention mechanism
JP4937395B2 (en) Feature vector generation apparatus, feature vector generation method and program
JP5014479B2 (en) Image search apparatus, image search method and program
Uricchio et al. An evaluation of nearest-neighbor methods for tag refinement
JP5518856B2 (en) Improved image recognition support device
CN112182264A (en) Method, device and equipment for determining landmark information and readable storage medium
CN111651596A (en) Text clustering method, text clustering device, server and storage medium
CN113094519B (en) Method and device for searching based on document
JP6017277B2 (en) Program, apparatus and method for calculating similarity between contents represented by set of feature vectors
JP5833499B2 (en) Retrieval device and program for retrieving content expressed by high-dimensional feature vector set with high accuracy
Du et al. Large-scale signature matching using multi-stage hashing
CN108170664B (en) Key word expansion method and device based on key words
CN115757896A (en) Vector retrieval method, device, equipment and readable storage medium
CN106557533B (en) Single-target multi-image joint retrieval method and device
CN114610859A (en) Product recommendation method, device and equipment based on content and collaborative filtering

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