CN108268904B - Picture identification method and device and electronic equipment - Google Patents
Picture identification method and device and electronic equipment Download PDFInfo
- Publication number
- CN108268904B CN108268904B CN201810171354.0A CN201810171354A CN108268904B CN 108268904 B CN108268904 B CN 108268904B CN 201810171354 A CN201810171354 A CN 201810171354A CN 108268904 B CN108268904 B CN 108268904B
- Authority
- CN
- China
- Prior art keywords
- picture
- analyzed
- sub
- determining
- image
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
- G06F18/2193—Validation; Performance evaluation; Active pattern learning techniques based on specific statistical tests
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a picture identification method, a picture identification device and electronic equipment. According to the invention, the analysis result of the picture to be analyzed can be directly determined without manual participation, so that the manual workload is reduced, and manual experience in the field of picture processing is not required.
Description
Technical Field
The invention relates to the field of image processing, in particular to a picture identification method and device and electronic equipment.
Background
At present, a mobile phone can enable people to take photos anytime and anywhere, and can enable people to share the processed photos anytime and anywhere after editing and modifying the taken photos by using software such as American picture and PS (Photoshop).
In the prior art, it is determined whether a picture is an original picture or a processed picture by manually analyzing whether the edge of the picture to be analyzed is clear or not and whether the gradual change is natural or not, but the method of manually analyzing whether the edge of the picture to be analyzed is clear or not and whether the gradual change is natural or not determines whether the picture is the original picture or the processed picture, so that the manual workload is large and the manual experience in the field of picture processing is required.
Disclosure of Invention
In view of this, the present invention provides a picture identification method, an apparatus and an electronic device, so as to solve the problems that manual work is heavy and manual experience in the picture processing field is required to determine whether a picture is an original picture or a processed picture by manually analyzing whether the edge of the picture to be analyzed is clear or whether gradual change is natural.
In order to solve the technical problems, the invention adopts the following technical scheme:
a picture identification method comprises the following steps:
segmenting a picture to be analyzed into a plurality of sub-region pictures;
acquiring the image characteristics of each subregion picture;
extracting attribute features which characterize the attribute of the corresponding subregion picture from the image features of each subregion picture;
determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture; and the analysis result is the result of whether picture processing is performed or not.
Preferably, determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture comprises:
and determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture and the neural network model.
Preferably, the generation process of the neural network model includes:
obtaining a sample picture; the sample pictures comprise a plurality of original picture samples and a plurality of picture samples after picture processing;
dividing each sample picture into a plurality of sub-pictures;
determining attribute characteristics of each sub-picture;
acquiring an initial neural network model;
training each neural layer in the initial neural network model based on the attribute characteristics of each sub-picture to obtain the parameters of each neural layer;
and obtaining the neural network model based on the parameters of each neural layer.
Preferably, when the analysis result of the picture to be analyzed includes a result processed by the picture, the method further includes:
and determining the original image which is not processed by the image and corresponds to the image to be analyzed.
Preferably, determining the original image which is not processed by the image and corresponds to the image to be analyzed includes:
determining intermediate pictures with the similarity to the picture to be analyzed being larger than a preset value and the similarity of each intermediate picture to the picture to be analyzed based on the picture to be analyzed and a preset search model;
determining an original picture in the intermediate picture based on the neural network model;
determining an original picture with the maximum similarity to the picture to be analyzed according to the similarity between each original picture and the picture to be analyzed;
and taking the determined original picture with the maximum similarity as the original picture which is not processed by the picture and corresponds to the picture to be analyzed.
A picture recognition apparatus comprising:
the first segmentation module is used for segmenting the picture to be analyzed into a plurality of sub-region pictures;
the first acquisition module is used for acquiring the image characteristics of each subregion picture;
the extraction module is used for extracting attribute features which characterize the attributes of the corresponding sub-region pictures from the image features of each sub-region picture;
the first determining module is used for determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture; and the analysis result is the result of whether picture processing is performed or not.
Preferably, the first determining module comprises:
and the first determining submodule is used for determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture and the neural network model.
Preferably, the method further comprises the following steps:
the second acquisition module is used for acquiring a sample picture; the sample pictures comprise a plurality of original picture samples and a plurality of picture samples after picture processing;
a second slicing module for slicing each sample picture into a plurality of sub-pictures;
a second determining module, configured to determine an attribute feature of each of the sub-pictures;
the third acquisition module is used for acquiring the initial neural network model;
the training module is used for training each neural layer in the initial neural network model based on the attribute characteristics of each sub-picture to obtain the parameters of each neural layer;
and the third determining module is used for obtaining the neural network model based on the parameters of each neural layer.
Preferably, the method further comprises the following steps:
and the fourth determining module is used for determining the original image which is not processed by the image and corresponds to the image to be analyzed when the analysis result of the image to be analyzed comprises the result processed by the image.
Preferably, the fourth determining module includes:
the second determining submodule is used for determining and obtaining intermediate pictures with the similarity to the picture to be analyzed being larger than a preset value and the similarity of each intermediate picture to the picture to be analyzed from a picture library based on the picture to be analyzed and a preset searching model;
the third determining sub-module is used for determining an original picture in the intermediate picture based on the neural network model;
the fourth determining submodule is used for determining the original picture with the maximum similarity to the picture to be analyzed according to the similarity between each original picture and the picture to be analyzed;
and the fifth determining submodule is used for taking the determined original picture with the maximum similarity as the original picture which is not processed by the picture and corresponds to the picture to be analyzed.
An electronic device, comprising: a memory and a processor;
wherein the memory is used for storing programs;
the processor is configured to invoke a program, wherein the program is configured to:
segmenting a picture to be analyzed into a plurality of sub-region pictures;
acquiring the image characteristics of each subregion picture;
extracting attribute features which characterize the attribute of the corresponding subregion picture from the image features of each subregion picture;
determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture; and the analysis result is the result of whether picture processing is performed or not.
Compared with the prior art, the invention has the following beneficial effects:
the invention provides a picture identification method, a picture identification device and electronic equipment. According to the invention, the analysis result of the picture to be analyzed can be directly determined without manual participation, so that the manual workload is reduced, and manual experience in the field of picture processing is not required.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flowchart of a method for identifying a picture according to the present invention;
FIG. 2 is a flowchart of another method for identifying pictures according to the present invention;
FIG. 3 is a flowchart of a method of another image recognition method according to the present invention;
fig. 4 is a schematic structural diagram of a picture recognition apparatus according to the present invention;
fig. 5 is a schematic structural diagram of another picture recognition device provided in the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, 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 invention.
The invention provides a picture identification method, which can comprise the following steps with reference to fig. 1:
s11, segmenting the picture to be analyzed into a plurality of sub-region pictures;
the picture to be analyzed is segmented by adopting a scanning window or a candidate window method.
The candidate window is determined by using information such as texture, edge, color, and the like in the image, for example, a target candidate window in which a face may exist in the image is detected by using a selective search strategy, and a plurality of candidate windows are searched.
If the picture to be analyzed is a picture of a cat, the picture is segmented, and a plurality of sub-region pictures of cat ears, cat legs, cat eyes and the like can be obtained.
S12, acquiring the image characteristics of each subregion picture;
the image features include brightness, edge, texture, color, and the like.
S13, extracting attribute features which characterize the attributes of the corresponding sub-region pictures from the image features of each sub-region picture;
the attribute can be a category, and attribute features which can represent the category are extracted from the image features.
And if the sub-region picture is a cat ear, extracting attribute features representing the cat ear from the image features.
The attribute feature can be extracted by adopting methods of nonlinear transformation, matrix dimension reduction, principal component analysis and the like.
It should be noted that the image features of each subregion picture are acquired to form a matrix.
S14, determining the analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture;
and the analysis result is the result of whether picture processing is performed or not.
Optionally, on the basis of this embodiment, step S14 may include:
and determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture and the neural network model.
Wherein the neural network model is pre-established. And inputting the attribute characteristics of the sub-region picture into the neural network model to obtain an analysis result of the picture to be analyzed.
In this embodiment, a picture to be analyzed is divided into a plurality of sub-region pictures, image features of each sub-region picture are obtained, attribute features which characterize attributes of the corresponding sub-region picture in the image features of each sub-region picture are extracted, and an analysis result of the picture to be analyzed is determined based on the attribute features of each sub-region picture. According to the invention, the analysis result of the picture to be analyzed can be directly determined without manual participation, so that the manual workload is reduced, and manual experience in the field of picture processing is not required.
Optionally, on the basis that the step S14 includes an embodiment that determines an analysis result of the picture to be analyzed based on the attribute features of each sub-region picture and a neural network model, referring to fig. 2, a generation process of the neural network model includes:
s21, obtaining a sample picture;
the sample pictures comprise a plurality of original picture samples and a plurality of picture samples after picture processing.
Specifically, the picture after the picture processing may be a PS picture,
s22, cutting each sample picture into a plurality of sub-pictures;
the splitting process is similar to the process of splitting the picture to be analyzed into a plurality of sub-region pictures.
S23, determining the attribute characteristics of each sub-picture;
s24, obtaining an initial neural network model;
s25, training each neural layer in the initial neural network model based on the attribute characteristics of each sub-picture to obtain the parameters of each neural layer;
in the neural network model, a plurality of hidden layers and a plurality of convolution cores in each layer are utilized to carry out convolution on input attribute characteristics, an output matrix in each layer is used as the input of the next layer, and finally a classifier is connected for classification, and parameters of each layer, including the number of nodes in each layer, weight values, bias items and the like, are modified through continuous training. And finally, training at each hidden layer to obtain some abstract feature maps. Where the reciprocal (or gradient) for each neural layer input can be found by back-propagating the derivative of that layer output (or next layer input), training the weights in all filters.
The feature map is a feature matrix obtained by performing convolution on the image and outputting the result of the calculation of each layer to the next layer through the calculation of the convolution layer.
The hidden layer refers to a neural layer in the neural network except for an input layer and an output layer. Because they do not accept external signals, also do not directly send signals to the outside, mainly used to solve the linear inseparable problem, the technical staff can't directly observe the situation of each layer, and then called as the hidden layer.
And S26, obtaining the neural network model based on the parameters of each neural layer.
After the neural network model is trained, the input image can be classified by using the neural network model.
In this embodiment, a method for generating a neural network model is provided, so that whether a picture is processed or not can be determined by using the generated neural network model.
Optionally, on the basis of any of the above embodiments, when the analysis result of the picture to be analyzed includes a result processed by the picture, the method further includes:
and determining the original image which is not processed by the image and corresponds to the image to be analyzed.
Optionally, on the basis of this embodiment, referring to fig. 3, the process of determining the original image corresponding to the picture to be analyzed and not processed by the picture may include:
s31, based on the picture to be analyzed and a preset search model, searching the picture library to obtain intermediate pictures with the similarity to the picture to be analyzed being larger than a preset value, and the similarity between each intermediate picture and the picture to be analyzed.
The preset search model is a model obtained by training based on a neural network algorithm, and data used by the preset search model is data in ImageNet.
ImageNet is a very large data set used in the field of deep learning images, and has a large number of labeled samples, such as labeled class-level labels and corresponding attributes. A search model is trained on this data set by a convolutional neural network.
The data used in this embodiment is a plurality of picture samples, where the information of each picture sample is manually labeled, and the information includes category features, content features included in the picture, and the like. Each picture sample needs to be sliced into sub-images and the attribute characteristics of each sub-image are determined.
Wherein, the class characteristics can be animals, plants, landscape, sea, etc. The content features may be inclusive of one large person and one small child, etc.
After a large amount of data is obtained, the attribute characteristics of each sub-image are input into a deep neural network model for training, and the deep neural network model for searching, namely a preset searching model, can be obtained.
After the preset search model is established, inputting the attribute characteristics of each sub-region picture of the picture to be analyzed into the preset search model, and obtaining the intermediate pictures with the similarity greater than a preset value with the picture to be analyzed and the similarity between each intermediate picture and the picture to be analyzed. The obtained intermediate picture may be an original picture, that is, a picture without picture processing, or may include a picture after picture processing, such as a PS picture. The original picture in the obtained intermediate picture is also labeled with a probability value of being the original picture, and the probability of being 99% may be the original picture.
It should be noted that the training in this embodiment is supervised learning, the error is propagated reversely by manually marking and using the error between the training value and the actual output value, and after the technician makes a manual judgment, the weight is manually adjusted to correct the deep neural network model.
S32, determining an original picture in the intermediate picture based on the neural network model;
specifically, the neural network model is used for identifying whether a picture is processed, that is, whether the picture is an original picture or a picture processed by the picture can be distinguished.
Further, the original picture in the intermediate picture obtained can be specified by using the neural network model.
S33, determining the original picture with the maximum similarity to the picture to be analyzed according to the similarity between each original picture and the picture to be analyzed;
specifically, the similarity is already obtained in the preset search model, and only the original picture with the highest similarity to the picture to be analyzed needs to be found.
And S34, taking the determined original picture with the maximum similarity as the original picture which is not processed by the picture and corresponds to the picture to be analyzed.
Specifically, the higher the similarity is, the more similar the original picture and the picture to be analyzed are, and the original picture is a picture which is not subjected to picture processing, so that the determined original picture with the maximum similarity can be used as the original picture which is not subjected to picture processing and corresponds to the picture to be analyzed.
It should be noted that, when the original pictures are searched again, the original picture with the highest similarity is preferentially searched, and if the similarities of the two original pictures and the picture to be analyzed are the same, the original picture with the highest probability of the original picture is selected as the original picture which is not processed by the picture and corresponds to the picture to be analyzed.
In this embodiment, an original image corresponding to the image processed by the image may be determined by using a preset search model and a neural network model, so that the PS image can be restored.
Optionally, on the basis of the embodiment of the picture identification method, another embodiment of the present invention provides a picture identification apparatus, and with reference to fig. 4, the picture identification apparatus may include:
the first segmentation module 11 is configured to segment a picture to be analyzed into a plurality of sub-region pictures;
a first obtaining module 12, configured to obtain an image feature of each sub-region picture;
the extraction module 13 is configured to extract attribute features, which characterize attributes of the corresponding sub-region picture, from the image features of each sub-region picture;
a first determining module 14, configured to determine an analysis result of the picture to be analyzed based on an attribute feature of each sub-region picture; and the analysis result is the result of whether picture processing is performed or not.
Optionally, on the basis of this embodiment, the first determining module includes:
and the first determining submodule is used for determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture and the neural network model.
In this embodiment, a picture to be analyzed is divided into a plurality of sub-region pictures, image features of each sub-region picture are obtained, attribute features which characterize attributes of the corresponding sub-region picture in the image features of each sub-region picture are extracted, and an analysis result of the picture to be analyzed is determined based on the attribute features of each sub-region picture. According to the invention, the analysis result of the picture to be analyzed can be directly determined without manual participation, so that the manual workload is reduced, and manual experience in the field of picture processing is not required.
It should be noted that, for the working processes of each module and sub-module in this embodiment, please refer to the corresponding description in the above embodiments, which is not described herein again.
Optionally, on the basis of the above embodiment in which the first determining module includes the first determining sub-module, the method further includes:
the second acquisition module is used for acquiring a sample picture; the sample pictures comprise a plurality of original picture samples and a plurality of picture samples after picture processing;
a second slicing module for slicing each sample picture into a plurality of sub-pictures;
a second determining module, configured to determine an attribute feature of each of the sub-pictures;
the third acquisition module is used for acquiring the initial neural network model;
the training module is used for training each neural layer in the initial neural network model based on the attribute characteristics of each sub-picture to obtain the parameters of each neural layer;
and the third determining module is used for obtaining the neural network model based on the parameters of each neural layer.
In this embodiment, a method for generating a neural network model is provided, so that whether a picture is processed or not can be determined by using the generated neural network model.
It should be noted that, for the working process of each module in this embodiment, please refer to the corresponding description in the above embodiments, which is not described herein again.
Optionally, on the basis of any of the above embodiments, the method further includes:
and the fourth determining module is used for determining the original image which is not processed by the image and corresponds to the image to be analyzed when the analysis result of the image to be analyzed comprises the result processed by the image.
Further, referring to fig. 5, the fourth determining module includes:
a second determining submodule 21, configured to determine, based on the picture to be analyzed and a preset search model, intermediate pictures whose similarity to the picture to be analyzed is greater than a preset value, and the similarity between each intermediate picture and the picture to be analyzed;
a third determining submodule 22, configured to determine an original picture in the intermediate picture based on the neural network model;
a fourth determining submodule 23, configured to determine, according to a similarity between each original picture and the picture to be analyzed, an original picture with a largest similarity to the picture to be analyzed;
and a fifth determining submodule 24, configured to use the determined original image with the largest similarity as the original image, which is not processed by the image, corresponding to the image to be analyzed.
In this embodiment, an original image corresponding to the image processed by the image may be determined by using a preset search model and a neural network model, so that the PS image can be restored.
It should be noted that, for the working processes of each module and sub-module in this embodiment, please refer to the corresponding description in the above embodiments, which is not described herein again.
Optionally, on the basis of the embodiments of the image recognition method and apparatus, another embodiment of the present invention provides an electronic device, including: a memory and a processor;
wherein the memory is used for storing programs;
the processor is configured to invoke a program, wherein the program is configured to:
segmenting a picture to be analyzed into a plurality of sub-region pictures;
acquiring the image characteristics of each subregion picture;
extracting attribute features which characterize the attribute of the corresponding subregion picture from the image features of each subregion picture;
determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture; and the analysis result is the result of whether picture processing is performed or not.
Further, the processor is configured to, when determining an analysis result of the picture to be analyzed based on the attribute feature of each sub-region picture, specifically:
and determining an analysis result of the picture to be analyzed based on the attribute characteristics of each sub-region picture and the neural network model.
Further, the processor is further configured to:
obtaining a sample picture; the sample pictures comprise a plurality of original picture samples and a plurality of picture samples after picture processing;
dividing each sample picture into a plurality of sub-pictures;
determining attribute characteristics of each sub-picture;
acquiring an initial neural network model;
training each neural layer in the initial neural network model based on the attribute characteristics of each sub-picture to obtain the parameters of each neural layer;
and obtaining the neural network model based on the parameters of each neural layer.
Further, when the analysis result of the picture to be analyzed includes a result processed by the picture, the processor is further configured to:
and determining the original image which is not processed by the image and corresponds to the image to be analyzed.
Further, when the processor is configured to determine the original image that is not processed by the image and corresponds to the image to be analyzed, the processor is specifically configured to:
determining intermediate pictures with the similarity to the picture to be analyzed being larger than a preset value and the similarity of each intermediate picture to the picture to be analyzed based on the picture to be analyzed and a preset search model;
determining an original picture in the intermediate picture based on the neural network model;
determining an original picture with the maximum similarity to the picture to be analyzed according to the similarity between each original picture and the picture to be analyzed;
and taking the determined original picture with the maximum similarity as the original picture which is not processed by the picture and corresponds to the picture to be analyzed.
In this embodiment, a picture to be analyzed is divided into a plurality of sub-region pictures, image features of each sub-region picture are obtained, attribute features which characterize attributes of the corresponding sub-region picture in the image features of each sub-region picture are extracted, and an analysis result of the picture to be analyzed is determined based on the attribute features of each sub-region picture. According to the invention, the analysis result of the picture to be analyzed can be directly determined without manual participation, so that the manual workload is reduced, and manual experience in the field of picture processing is not required.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. 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 invention. Thus, the present invention 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 (7)
1. A picture recognition method is characterized by comprising the following steps:
segmenting a picture to be analyzed into a plurality of sub-region pictures;
acquiring the image characteristics of each subregion picture;
extracting attribute features which characterize the attributes of the corresponding sub-region pictures from the image features of each sub-region picture, wherein the attribute features are attribute features which characterize categories;
inputting the attribute characteristics of each subregion picture into a pre-established neural network model to obtain the analysis result of the analysis picture; wherein the analysis result is the result of whether picture processing is performed or not;
wherein, the generation process of the neural network model comprises the following steps:
obtaining a sample picture; the sample pictures comprise a plurality of original picture samples and a plurality of picture samples after picture processing;
dividing each sample picture into a plurality of sub-pictures;
determining attribute characteristics of each sub-picture;
acquiring an initial neural network model;
training each neural layer in the initial neural network model based on the attribute characteristics of each sub-picture to obtain the parameters of each neural layer;
and obtaining the neural network model based on the parameters of each neural layer.
2. The picture recognition method according to claim 1, wherein when the analysis result of the picture to be analyzed includes a result of picture processing, the method further comprises:
and determining the original image which is not processed by the image and corresponds to the image to be analyzed.
3. The method according to claim 2, wherein determining the original image corresponding to the picture to be analyzed and not processed by the picture comprises:
determining intermediate pictures with the similarity to the picture to be analyzed being larger than a preset value and the similarity of each intermediate picture to the picture to be analyzed based on the picture to be analyzed and a preset search model;
determining an original picture in the intermediate picture based on the neural network model;
determining an original picture with the maximum similarity to the picture to be analyzed according to the similarity between each original picture and the picture to be analyzed;
and taking the determined original picture with the maximum similarity as the original picture which is not processed by the picture and corresponds to the picture to be analyzed.
4. An image recognition apparatus, comprising:
the first segmentation module is used for segmenting the picture to be analyzed into a plurality of sub-region pictures;
the first acquisition module is used for acquiring the image characteristics of each subregion picture;
the extraction module is used for extracting attribute features which characterize the attributes of the corresponding sub-region pictures from the image features of each sub-region picture, wherein the attribute features are attribute features which characterize categories;
the first determining module is used for inputting the attribute characteristics of each subregion picture into a pre-established neural network model to obtain the analysis result of the analysis picture; wherein the analysis result is the result of whether picture processing is performed or not;
the second acquisition module is used for acquiring a sample picture; the sample pictures comprise a plurality of original picture samples and a plurality of picture samples after picture processing;
a second slicing module for slicing each sample picture into a plurality of sub-pictures;
a second determining module, configured to determine an attribute feature of each of the sub-pictures;
the third acquisition module is used for acquiring the initial neural network model;
the training module is used for training each neural layer in the initial neural network model based on the attribute characteristics of each sub-picture to obtain the parameters of each neural layer;
and the third determining module is used for obtaining the neural network model based on the parameters of each neural layer.
5. The picture recognition device according to claim 4, further comprising:
and the fourth determining module is used for determining the original image which is not processed by the image and corresponds to the image to be analyzed when the analysis result of the image to be analyzed comprises the result processed by the image.
6. The picture recognition device according to claim 5, wherein the fourth determination module comprises:
the second determining submodule is used for determining and obtaining intermediate pictures with the similarity to the picture to be analyzed being larger than a preset value and the similarity of each intermediate picture to the picture to be analyzed from a picture library based on the picture to be analyzed and a preset searching model;
the third determining sub-module is used for determining an original picture in the intermediate picture based on the neural network model;
the fourth determining submodule is used for determining the original picture with the maximum similarity to the picture to be analyzed according to the similarity between each original picture and the picture to be analyzed;
and the fifth determining submodule is used for taking the determined original picture with the maximum similarity as the original picture which is not processed by the picture and corresponds to the picture to be analyzed.
7. An electronic device, comprising: a memory and a processor;
wherein the memory is used for storing programs;
the processor is configured to invoke a program, wherein the program is configured to:
segmenting a picture to be analyzed into a plurality of sub-region pictures;
acquiring the image characteristics of each subregion picture;
extracting attribute features which characterize the attributes of the corresponding sub-region pictures from the image features of each sub-region picture, wherein the attribute features are attribute features which characterize categories;
inputting the attribute characteristics of each subregion picture into a pre-established neural network model to obtain the analysis result of the analysis picture; wherein the analysis result is the result of whether picture processing is performed or not;
wherein, the generation process of the neural network model comprises the following steps:
obtaining a sample picture; the sample pictures comprise a plurality of original picture samples and a plurality of picture samples after picture processing;
dividing each sample picture into a plurality of sub-pictures;
determining attribute characteristics of each sub-picture;
acquiring an initial neural network model;
training each neural layer in the initial neural network model based on the attribute characteristics of each sub-picture to obtain the parameters of each neural layer;
and obtaining the neural network model based on the parameters of each neural layer.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810171354.0A CN108268904B (en) | 2018-03-01 | 2018-03-01 | Picture identification method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810171354.0A CN108268904B (en) | 2018-03-01 | 2018-03-01 | Picture identification method and device and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108268904A CN108268904A (en) | 2018-07-10 |
| CN108268904B true CN108268904B (en) | 2022-03-25 |
Family
ID=62774416
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810171354.0A Active CN108268904B (en) | 2018-03-01 | 2018-03-01 | Picture identification method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108268904B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108921126B (en) * | 2018-07-20 | 2019-07-23 | 北京开普云信息科技有限公司 | A kind of automatic identification signature stamp or the method and device of handwritten signature |
| CN113111888B (en) * | 2021-04-15 | 2024-04-26 | 广州图匠数据科技有限公司 | Picture discrimination method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103258186A (en) * | 2013-03-17 | 2013-08-21 | 北京工业大学 | Integrated face recognition method based on image segmentation |
| CN105469087A (en) * | 2015-07-13 | 2016-04-06 | 百度在线网络技术（北京）有限公司 | Method for identifying clothes image, and labeling method and device of clothes image |
| CN105808610A (en) * | 2014-12-31 | 2016-07-27 | 中国科学院深圳先进技术研究院 | Internet picture filtering method and device |
| CN106156777A (en) * | 2015-04-23 | 2016-11-23 | 华中科技大学 | Textual image detection method and device |
| CN106407991A (en) * | 2016-09-14 | 2017-02-15 | 北京市商汤科技开发有限公司 | Image attribute identification method and system and related network training method and system |
| CN107622489A (en) * | 2017-10-11 | 2018-01-23 | 广东工业大学 | A kind of distorted image detection method and device |
-
2018
- 2018-03-01 CN CN201810171354.0A patent/CN108268904B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103258186A (en) * | 2013-03-17 | 2013-08-21 | 北京工业大学 | Integrated face recognition method based on image segmentation |
| CN105808610A (en) * | 2014-12-31 | 2016-07-27 | 中国科学院深圳先进技术研究院 | Internet picture filtering method and device |
| CN106156777A (en) * | 2015-04-23 | 2016-11-23 | 华中科技大学 | Textual image detection method and device |
| CN105469087A (en) * | 2015-07-13 | 2016-04-06 | 百度在线网络技术（北京）有限公司 | Method for identifying clothes image, and labeling method and device of clothes image |
| CN106407991A (en) * | 2016-09-14 | 2017-02-15 | 北京市商汤科技开发有限公司 | Image attribute identification method and system and related network training method and system |
| CN107622489A (en) * | 2017-10-11 | 2018-01-23 | 广东工业大学 | A kind of distorted image detection method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108268904A (en) | 2018-07-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106920229B (en) | Automatic detection method and system for image fuzzy area | |
| CN105808610B (en) | Internet picture filtering method and device | |
| CN103824053B (en) | The sex mask method and face gender detection method of a kind of facial image | |
| CN108562589B (en) | Method for detecting surface defects of magnetic circuit material | |
| CN108960245B (en) | Tire mold character detection and recognition method, device, equipment and storage medium | |
| CN109284733B (en) | Shopping guide negative behavior monitoring method based on yolo and multitask convolutional neural network | |
| CN109583483B (en) | Target detection method and system based on convolutional neural network | |
| WO2021036059A1 (en) | Image conversion model training method, heterogeneous face recognition method, device and apparatus | |
| US9129191B2 (en) | Semantic object selection | |
| CN109767422A (en) | Pipe detection recognition methods, storage medium and robot based on deep learning | |
| US20150170006A1 (en) | Semantic object proposal generation and validation | |
| CN104537341B (en) | Face picture information getting method and device | |
| CN106126585B (en) | The unmanned plane image search method combined based on quality grading with perceived hash characteristics | |
| CN107153838A (en) | A kind of photo automatic grading method and device | |
| WO2022089170A1 (en) | Caption area identification method and apparatus, and device and storage medium | |
| CN110009622B (en) | Display panel appearance defect detection network and defect detection method thereof | |
| CN108764084A (en) | Video classification methods based on spatial domain sorter network and the time domain network integration | |
| CN106408086A (en) | Deep learning neural network processing method and deep learning neural network processing system for image optimization | |
| CN108268904B (en) | Picture identification method and device and electronic equipment | |
| CN111126143A (en) | Deep learning-based exercise judgment guidance method and system | |
| CN114187595A (en) | Document layout recognition method and system based on fusion of visual features and semantic features | |
| CN114005019B (en) | Method for identifying flip image and related equipment thereof | |
| WO2018120575A1 (en) | Method and device for identifying main picture in web page | |
| CN112560925A (en) | Complex scene target detection data set construction method and system | |
| CN109165551B (en) | Expression recognition method for adaptively weighting and fusing significance structure tensor and LBP characteristics |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |