CN109274945A - A kind of adaptive method and system for carrying out the reduction of image true color - Google Patents
A kind of adaptive method and system for carrying out the reduction of image true color Download PDFInfo
- Publication number
- CN109274945A CN109274945A CN201811143926.0A CN201811143926A CN109274945A CN 109274945 A CN109274945 A CN 109274945A CN 201811143926 A CN201811143926 A CN 201811143926A CN 109274945 A CN109274945 A CN 109274945A
- Authority
- CN
- China
- Prior art keywords
- image
- training
- true color
- shot
- reference object
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 230000003044 adaptive effect Effects 0.000 title abstract description 5
- 238000012549 training Methods 0.000 claims abstract description 196
- 238000001228 spectrum Methods 0.000 claims abstract description 82
- 238000003384 imaging method Methods 0.000 claims abstract description 42
- 239000000523 sample Substances 0.000 claims abstract description 23
- 230000003287 optical effect Effects 0.000 claims abstract description 17
- 230000007613 environmental effect Effects 0.000 claims description 32
- 238000012217 deletion Methods 0.000 claims description 17
- 230000037430 deletion Effects 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 8
- 230000003595 spectral effect Effects 0.000 abstract description 14
- 230000006870 function Effects 0.000 description 12
- 238000012544 monitoring process Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 8
- 239000003086 colorant Substances 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 4
- 238000011835 investigation Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 206010039203 Road traffic accident Diseases 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000005316 response function Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/10—Cameras or camera modules comprising electronic image sensors; Control thereof for generating image signals from different wavelengths
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
- H04N23/80—Camera processing pipelines; Components thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
- Facsimile Image Signal Circuits (AREA)
Abstract
The application provides a kind of adaptive method and system for carrying out the reduction of image true color, and the grey scale pixel value of image and the corresponding relationship of image true color information are shot under ambient light spectrum information condition this method comprises: being trained to obtain according to the grey scale pixel value of reference object and the true color information of reference object in training set；Receive the current taken image of imaging device and the current environment spectral information with the spectroscopic probe head of imaging device optical axis center balance；Analysis current taken image obtains the gray value of each pixel of current taken image；The pixel grey scale offset of current taken image is obtained according to the corresponding relationship of the grey scale pixel value and image true color information that shoot image under current context information；Original image is gone back according to what the gray value of each pixel of pixel grey scale offset and current taken image of current taken image obtained current taken image.The present invention can fast and accurately restore image color.
Description
Technical Field
The present application relates to the field of image processing technologies, and in particular, to a method and a system for adaptively performing real color restoration on an image.
Background
The security monitoring is very important for maintaining social stability, enhancing urban public security management and improving the safety and the happiness of people's life. In various interactive scenes, the security system can record and store scene events in real time through images, so that the scene conditions at that time can be conveniently and subsequently checked and restored, and a large amount of original data and evidences are provided for pursuing illegal criminal behaviors and the like.
However, most current security systems employ conventional imaging devices, i.e., fixed filters and band response functions to generate RGB images or videos. Under sufficient light, an RGB image or video generated by a traditional imaging system accords with the imaging rule of human eyes and is suitable for human eye observation, but under the condition of dim light or unbalanced illumination, the condition of image color distortion easily occurs, the defect that the real color of an object in a scene cannot be really recorded has huge influence in actual life, and if the color of a vehicle, clothes of a suspect or the appearance of a specific scene is identified, the image and the video with color distortion cause the error of a locked target or cannot be clearly distinguished, so that the subsequent investigation work is delayed.
As can be known from optical knowledge, light in a specific spectrum band has specific spectral information, and different ambient lights have different influences on a scene, and only an image obtained through a light compensation strategy is imaged and cannot truly reflect the situation of the current scene, or cannot clearly reflect the situation of the current scene, so how to provide a scheme for restoring an image of a real scene by performing image processing in real time and adaptively according to different scenes is a technical problem to be solved in the art.
Disclosure of Invention
The application aims to provide a method and a system for carrying out image true color restoration in a self-adaptive manner, and solve the technical problem that the image true color restoration cannot be carried out in a real-time self-adaptive manner according to a scene in the prior art.
In order to achieve the above object, the present application provides a method for adaptively performing true color restoration of an image, including:
presetting a reference object, and acquiring true color information of the reference object; capturing an image with the reference object as a training captured image with an imaging device;
detecting environmental spectrum information when the training shot image is shot as training environmental spectrum information through a spectrum probe balanced with the center of the optical axis of the imaging equipment;
receiving the training shot image, the training environment spectrum information and the true color information of the reference object to form a training set; analyzing the training shot image to obtain the pixel gray value of the reference object in the training shot image;
training according to the pixel gray value of the reference object in the training set and the true color information of the reference object to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the shot image under the condition of environmental spectrum information;
shooting a target image by using the imaging equipment, and detecting real-time environment spectrum information when the target image is shot by using the spectrum probe balanced with the center of the optical axis of the imaging equipment;
receiving the target image and the real-time environment spectrum information which are shot currently by the imaging equipment; analyzing the currently shot target image to obtain the gray value of each pixel of the target image;
obtaining a pixel gray compensation value of the target image according to the corresponding relation between the pixel gray value of the target image and the true color information of the image under the current environment information;
and obtaining a restored image of the currently shot target image according to the pixel gray compensation value of the target image and the gray value of each pixel of the target image.
Optionally, wherein the method further comprises:
in the training set, obtaining a restored image of the reference object according to the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the environmental spectrum information and the pixel gray value of the reference object;
obtaining a loss value of image restoration according to the restored image of the reference object and the true color information of the reference object; when the loss value of the image restoration is smaller than or equal to a preset loss value threshold, storing the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectrum information condition;
and when the loss value of the image restoration is greater than the preset loss value threshold, generating an image of the restored image of the reference object and the loss function of the true color information.
Optionally, wherein the method further comprises:
in the training set, receiving an instruction for processing data in the training set, and analyzing the instruction to obtain a data identifier of a training image indicating deletion;
searching and deleting the data of the training image indicating deletion according to the data identification of the training image indicating deletion;
and obtaining the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectrum information condition according to the residual data in the training set.
Optionally, wherein the loss function of the restored image and the true color information is:
wherein,
pi(x, y) represents the image gray value of the reconstructed restored image at the ith pixel point, tiAnd (x, y) represents the image gray value of the real true color image at the ith pixel point.
Optionally, wherein the method further comprises:
in the training set, receiving a newly added training shot image, newly added training environment spectrum information and updated true color information of a reference object in the newly added training shot image to update data of the training set;
and training by using the updated data in the training set to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the updated environmental spectrum information.
In another aspect, the present invention further provides a system for adaptively performing true color restoration on an image, including: the system comprises an image training data collector, an image data training analyzer, a real-time image data receiver, a real-time image compensation analyzer and a real-time image restorer; wherein,
the image training data collector is connected with the image data training analyzer and used for receiving training shot images of imaging equipment, training environment spectrum information of a spectrum probe balanced with the center of an optical axis of the imaging equipment and true color information of reference objects in the training shot images to form a training set; analyzing the training shot image to obtain the pixel gray value of the reference object in the training shot image;
the image data training analyzer is connected with the image training data collector and the real-time image data receiver and is used for training according to the pixel gray value of the reference object in the training set and the true color information of the reference object to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the shot image under the condition of environmental spectrum information;
the real-time image data receiver is connected with the image data training analyzer and the real-time image compensation analyzer and is used for receiving a current shot image of the imaging equipment and current environment spectrum information of the spectrum probe balanced with the center of the optical axis of the imaging equipment; analyzing the current shot image to obtain the gray value of each pixel of the current shot image;
the real-time image compensation analyzer is connected with the real-time image data receiver and the real-time image restorer and is used for obtaining a pixel gray level compensation value of the current shot image according to the corresponding relation between the pixel gray level value of the shot target image under the current environment information and the true color information of the image;
and the real-time image restorer is connected with the real-time image compensation analyzer and is used for obtaining a restored image of the target image according to the pixel gray level compensation value of the target image and the gray level value of each pixel of the target image.
Optionally, wherein the system further comprises: an image data training evaluator coupled to the image data training analyzer for:
in the training set, obtaining a restored image of the reference object according to the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the environmental spectrum information and the pixel gray value of the reference object;
obtaining a loss value of image restoration according to the restored image of the reference object and the true color information of the reference object; when the loss value of the image restoration is smaller than or equal to a preset loss value threshold, storing the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectrum information condition;
and when the loss value of the image restoration is greater than the preset loss value threshold, generating an image of the restored image of the reference object and the loss function of the true color information.
Optionally, wherein the system further comprises: a training image data processor coupled to the image data training evaluator to:
receiving an instruction for processing the data in the training set, and analyzing the instruction to obtain a data identifier of a training image which indicates deletion;
searching and deleting the data of the training image indicating deletion according to the data identification of the training image indicating deletion;
and obtaining the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectrum information condition according to the residual data in the training set.
Optionally, wherein the loss function of the restored image and the true color information is:
wherein,
pi(x, y) represents the image gray value of the reconstructed restored image at the ith pixel point, tiAnd (x, y) represents the image gray value of the real true color image at the ith pixel point.
Optionally, wherein the system further comprises: a training image data updater connected to the image data training analyzer and configured to:
receiving a newly added training shot image, newly added training environment spectrum information and updated true color information of a reference object in the newly added training shot image to update data of the training set;
and training by using the updated data in the training set to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the updated environmental spectrum information.
The beneficial effect that this application realized is as follows:
(1) according to the method and the system for self-adaptively carrying out image true color restoration, the first part is composed of the imaging device and the spectrum probe, the scene image can be shot in real time, the spectrum characteristics of the ambient light can be sensed, the second part is a trained neural network, the true color of the shot image is restored at the millisecond level in a self-adaptive mode according to the image and the spectrum characteristics transmitted by the first part, and the problem of image color distortion is solved. The network training process can train different amounts of ambient light according to actual needs, and after the network training is completed, the spectral characteristics of different ambient light can be detected by the probe, so that the purpose of quickly and accurately restoring the image color can be achieved.
(2) According to the method and the system for adaptively carrying out image true color restoration, the authenticity of the corresponding relation between the pixel gray value of the shot image obtained by data training in the training set and the image true color information is judged and evaluated through the loss function, the accuracy of adaptively restoring the shot image is improved in real time, and the image restoration effect is further improved.
(3) According to the method and the system for adaptively restoring the true color of the image, the authenticity of the corresponding relation between the pixel gray value of the shot image obtained by data training in the training set and the true color information of the image is updated in real time according to the training image data updated at any time, and the real-time accuracy of adaptively restoring the shot image is ensured.
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 described below, it is obvious that the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art according to the drawings.
FIG. 1 is a flowchart illustrating a first method for adaptively performing true color restoration on an image according to an embodiment of the present invention;
FIG. 2 is a schematic flowchart illustrating a schematic flow chart of a method for adaptively performing true color restoration on an image according to an embodiment of the present invention;
FIG. 3 is a schematic flowchart illustrating an embodiment of an image restoration processing method for adaptively performing true color restoration on an image shown in FIG. 1 according to the present invention;
FIG. 4 is a flowchart illustrating a second method for adaptively performing true color restoration on an image according to an embodiment of the present invention;
FIG. 5 is a flowchart illustrating a third method for adaptively performing true color restoration on an image according to an embodiment of the present invention;
FIG. 6 is a flowchart illustrating a fourth method for adaptively performing true color restoration on an image according to an embodiment of the present invention;
FIG. 7 is a schematic structural diagram of a first system for adaptively performing true color restoration on an image according to an embodiment of the present invention;
FIG. 8 is a diagram illustrating a second system for adaptively performing true color restoration on an image according to an embodiment of the present invention;
FIG. 9 is a schematic structural diagram of a third system for adaptively performing true color restoration on an image according to an embodiment of the present invention;
fig. 10 is a schematic diagram of a fourth system for adaptively performing true color restoration on an image according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present application are 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 some, but not all, embodiments of the present application. 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.
Examples
As shown in fig. 1, fig. 2 and fig. 3, fig. 1 is a schematic flow chart of a first method for adaptively performing true color restoration of an image according to the present embodiment; FIG. 2 is a schematic flowchart illustrating a schematic flow chart of a method for adaptively performing true color restoration on an image according to an embodiment of the present invention; fig. 3 is a schematic flow chart illustrating a schematic flow of an image restoration processing method for adaptively performing true color restoration on an image according to an embodiment of the present invention. The influence of several kinds of common ambient light on the true color of the object in real life is found, the deep learning network is built, the law is learned, and the true color of the object in imaging can be restored in real time according to the learned law. The self-adaptive true color restoration monitoring system is an effective idea for greatly improving the security accuracy and reliability. The method comprises the following steps:
step 101, presetting a reference object and acquiring true color information of the reference object; capturing an image with a reference object as a training captured image with an imaging device; and detecting environmental spectrum information when the training shot image is shot as training environmental spectrum information through a spectrum probe balanced with the center of an optical axis of the imaging equipment.
102, receiving a training shot image of imaging equipment and training environment spectrum information of a spectrum probe balanced with the center of an optical axis of the imaging equipment, and forming a training set by true color information of a reference object in the training shot image; and analyzing the training shot image to obtain the pixel gray value of the reference object in the training shot image.
The spectral information of the images shot under various environmental conditions is collected, and a set of training data is formed by the spectral information and the true color information of objects (such as preset human clothes) preset in the images under the conditions. The real corresponding relation between the image and the spectrum under the condition can be embodied through a proper amount of training data, so that the real corresponding relation is used as a compensation difference value between the image and the real image needing to be restored under the environment condition, and the restored image close to the real image is restored and obtained.
The idea of combining the traditional imaging device with a sensitive probe can obtain the spectral characteristics of the shot image and the current ambient light compared with the traditional imaging device. The number of sensitive probes and conventional imaging devices can be two, three or more.
And 103, training according to the pixel gray value of the reference object in the training set and the true color information of the reference object to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the shot image under the condition of environmental spectrum information.
Under different environmental conditions, the image obtained by shooting has a direct relation with the current spectrum, so that the color compensation between the shot image and the real image is influenced, the corresponding relation between the image, the spectrum and the real image is obtained through training of the training set, and the corresponding relation is applied to the compensation value of the image and the real image in other real-time image shooting scenes, so that a restored image close to the real image is restored and obtained.
The corresponding relationship between the pixel gray value of the image shot under the condition that the convolutional neural network and the loss function obtain the environmental spectral information and the true color information of the image is only a simple demonstration, and other convolutional neural networks, loss functions, convolutional neural networks and loss function algorithms can be converted into other algorithms, so long as the algorithms belong to the protection range of the implementation of restoring true color images by combining the image information and the spectral characteristics.
And 104, shooting a target image by using the imaging equipment, and detecting real-time environment spectrum information when the target image is shot by using a spectrum probe balanced with the center of an optical axis of the imaging equipment.
105, receiving a current shot image of the imaging equipment and current environment spectrum information of a spectrum probe balanced with the center of an optical axis of the imaging equipment; and analyzing the current shot image to obtain the gray value of each pixel of the current shot image.
In order to obtain more real-time image shooting information, the spectrum detection range of the probe simultaneously comprises ultraviolet light, visible light and infrared light, for example, 10 nm-1000 nm can be selected, and the detection angle is 360 degrees. The image is shot in real time through the shooting device, information of each pixel in the image or information of the set image is obtained, and basic data are provided for subsequent image restoration.
And 106, obtaining a pixel gray compensation value of the current shot image according to the corresponding relation between the pixel gray value of the shot image under the current environment information and the true color information of the image.
And according to the corresponding relation between the spectrum of the shot image and the true color image information under the specific environmental condition obtained by training the training data in the training set in the step, obtaining a gray compensation value which needs to be carried out on the pixel on the current shot image, and further compensating the real-time shot image through the gray compensation value to obtain a restored image which is close to a true color image.
And step 107, obtaining a restored image of the current shot image according to the pixel gray compensation value of the current shot image and the gray value of each pixel of the current shot image.
For example, under normal conditions, light is sufficient in a hall of an ATM cash dispenser, and the situation of serious color distortion cannot occur. However, in the recent robbery event of the ATM automatic teller machine, the suspect often destroys the lighting equipment in the hall in advance, so that the monitoring system cannot capture the real clothing color of the suspect, thereby affecting the investigation and the case-breaking of the public security organization. Under this condition, the self-adaptation true color control that this patent provided restores the system effect outstanding, and the built-in probe of system gathers the spectral feature of current ambient light in real time, and in the twinkling of an eye that the suspect destroys the lighting environment in hall, the probe is sharp catches the spectral feature of the environment after the change, and to the different spectral feature of input, the convolutional neural network of having trained right will restore out the true color of object at once, and according to the throughput of present deep neural network to the image, the speed of restoring of a picture is very fast, hardly has the time delay to appear.
Furthermore, in order to ensure traffic safety, a plurality of monitoring devices are arranged on the highway. In the daytime, the traditional monitoring equipment can basically and accurately capture the information of the passing vehicles, but in the night, dim ambient light can seriously affect the traditional monitoring equipment, so that the colors of the vehicles in the images are distorted, and the appearance information of the vehicles cannot be correctly determined. However, statistics related to traffic control departments show that the probability of traffic accidents at night is 1.5 times higher than that during the day, and 55% of traffic accidents occur at night. Therefore, the traditional monitoring equipment of the expressway is replaced by the self-adaptive true-color restoration monitoring system, the appearance colors of passing vehicles are accurately restored, the acquisition degree of monitoring information at night can be greatly improved, the follow-up tracking investigation is facilitated, and the problem that the colors of the vehicles at night are difficult to identify is solved.
Moreover, in a shopping mall with dense people flow or a case that a child is lost or induced to turn frequently occurs, the information provided by the monitoring equipment is of great importance in the process of catching a suspect. However, there are many areas in the market, the illumination condition in places such as exit, stair corner is not good enough, under the inhomogeneous or dim light condition of illumination, traditional supervisory equipment hardly catches lost children or suspect's true dress characteristic accurately. Based on the current situation, the self-adaptive true-color restoration monitoring system is installed in an area with dense people flow, so that the dressing characteristics of the lost children or the suspects can be accurately positioned, and accurate guide information is provided for subsequent positioning and tracking.
And the flow of people in railway stations and airports is huge, the incidence rate of cases such as theft, robbery and abduction is high, and especially at night, the traditional monitoring equipment cannot accurately capture the dressing color of people, thereby greatly influencing the further tracking of law enforcement officers. Therefore, the self-adaptive true color restoration system can be installed at railway stations and airports, the wearing of people in a monitoring range and the colors of personal belongings can be accurately restored, and investigation and tracking of law enforcement personnel are effectively assisted.
In the above scenes, the method for adaptively restoring true colors of images in the embodiment can be used to achieve the effect of restoring the images at that time, thereby helping operators to find accurate perpetrators and the like.
Optionally, as shown in fig. 4, which is a schematic flow chart of a second method for adaptively performing true color restoration on an image in this embodiment, different from fig. 1, the method further includes the following steps:
step 401, in the training set, obtaining a restored image of the reference object according to the correspondence between the pixel gray-scale value of the captured image and the true color information of the image under the condition of the environmental spectrum information and the pixel gray-scale value of the reference object.
Step 402, obtaining a loss value of image restoration according to the restored image of the reference object and the true color information of the reference object; and when the loss value of the image restoration is smaller than or equal to a preset loss value threshold, storing the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectral information condition.
And step 403, when the loss value of the image restoration is greater than a preset loss value threshold, generating a restored image of the reference object and an image of a loss function of the true color information.
Optionally, the loss function of the restored image and the true color information is:
wherein,
pi(x, y) represents the image gray value of the reconstructed restored image at the ith pixel point, tiAnd (x, y) represents the image gray value of the real true color image at the ith pixel point.
In some optional embodiments, as shown in fig. 5, which is a schematic flow chart of a third method for adaptively performing true color restoration on an image in this embodiment, different from fig. 4, the method includes the following steps:
step 501, in the training set, receiving an instruction for processing data in the training set, and analyzing the instruction to obtain a data identifier of a training image indicating deletion.
Step 502, searching and deleting the data of the training image indicating deletion according to the data identification of the training image indicating deletion.
Step 503, obtaining the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the environmental spectrum information according to the residual data in the training set.
Optionally, as shown in fig. 6, which is a schematic flow chart of a fourth method for adaptively performing true color restoration on an image according to an embodiment of the present invention, different from that in fig. 1, the method further includes:
step 601, in the training set, receiving the newly added training shot image, the newly added training environment spectrum information and the updated true color information of the reference object in the newly added training shot image to update the data of the training set.
Step 602, training by using the updated data in the training set to obtain a corresponding relationship between the pixel gray value of the captured image and the true color information of the image under the updated environment spectrum information condition.
Fig. 7 is a schematic structural diagram of a system 700 for performing an adaptive true color restoration of an image according to this embodiment, where the system is configured to perform the above method for performing an adaptive true color restoration of an image, and includes: an image training data collector 701, an image data training analyzer 702, a real-time image data receiver 703, a real-time image compensation parser 704 and a real-time image restorer 705.
The image training data collector 701 is connected with the image data training analyzer 702 and is used for receiving a training shot image of the imaging device, training environment spectrum information of a spectrum probe balanced with the center of an optical axis of the imaging device and true color information of a reference object in the training shot image to form a training set; and analyzing the training shot image to obtain the pixel gray value of the reference object in the training shot image.
Presetting a reference object before an image training data collector, and acquiring true color information of the reference object; capturing an image with a reference object as a training captured image with an imaging device; and detecting environmental spectrum information when the training shot image is shot as training environmental spectrum information through a spectrum probe balanced with the center of an optical axis of the imaging equipment. The corresponding relation between the pixel gray value of the shot image and the true color information of the image obtained by training in the system for self-adaptively performing the true color restoration of the image by utilizing the preset training data can be applied to a general shot image, so that the restored image of the general shot image can be quickly and accurately obtained.
The image data training analyzer 702 is connected to the image training data collector 701 and the real-time image data receiver 703, and configured to perform training according to the pixel grayscale value of the reference object in the training set and the true color information of the reference object to obtain a corresponding relationship between the pixel grayscale value of the captured image and the true color information of the image under the condition of the environmental spectral information.
The real-time image data receiver 703 is connected with the image data training analyzer 702 and the real-time image compensation analyzer 704, and is configured to receive a currently-captured image of the imaging device and current environmental spectrum information of the spectrum probe balanced with the optical axis center of the imaging device; and analyzing the current shot image to obtain the gray value of each pixel of the current shot image.
And the real-time image compensation analyzer 704 is connected with the real-time image data receiver 703 and the real-time image restorer 705, and is configured to obtain a pixel gray compensation value of the currently captured image according to a corresponding relationship between a pixel gray value of the captured image under the current environment information and the true color information of the image.
The real-time image restorer 705 is connected to the real-time image compensation analyzer 704, and is configured to obtain a restored image of the current captured image according to the pixel gray compensation value of the current captured image and the gray value of each pixel of the current captured image.
In some optional embodiments, as shown in fig. 8, a schematic structural diagram of a system 800 for performing image true color restoration for a second adaptive mode in this implementation, different from fig. 7, further includes: an image data training evaluator 801, connected to the image data training analyzer 702, for: in the training set, obtaining a restored image of the reference object according to the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the environmental spectrum information and the pixel gray value of the reference object; obtaining a loss value of image restoration according to the restored image of the reference object and the true color information of the reference object; when the loss value of the image restoration is smaller than or equal to a preset loss value threshold, storing the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the environmental spectrum information; and when the loss value of the image restoration is larger than a preset loss value threshold value, generating a restored image of the reference object and an image of a loss function of the true color information.
Optionally, the loss function of the restored image and the true color information is:
wherein,
pi(x, y) represents the image gray value of the reconstructed restored image at the ith pixel point, ti(x, y) represents the image gray value of the real true color image at the ith pixel point, and when the two images are the closest, the loss can reach the minimum.
The network adopts a self-adaptive training mode, and the reconstruction capability learned by the network training set is stronger and stronger along with more and more images passing through the network training set, so that a true color image and the true color of an object are reconstructed for the spectral characteristics of the shot image and the ambient light input under any ambient light.
In some optional embodiments, as shown in fig. 9, a schematic structural diagram of a system 900 for performing image true color restoration for a third adaptation in this implementation, different from fig. 8, further includes: a training image data processor 901 connected to the image data training evaluator 801 for: receiving an instruction for processing data in the training set, and analyzing the instruction to obtain a data identifier of the training image which is indicated to be deleted; searching and deleting the data of the training images indicating deletion according to the data identification of the training images indicating deletion; and obtaining the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the environmental spectrum information according to the residual data in the training set.
In some optional embodiments, as shown in fig. 10, a schematic structural diagram of a system 1000 for performing image true color restoration for a fourth adaptation in this implementation, different from fig. 7, further includes: a training image data updater 1001 connected to the image data training analyzer 702, for: receiving the newly added training shot image, the newly added training environment spectrum information and the updated true color information of the reference object in the newly added training shot image to update the data of the training set; and training by using the updated data in the training set to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the updated environmental spectrum information.
While the preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all alterations and modifications as fall within the scope of the application. It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.
Claims (10)
1. A method for adaptively performing true color restoration of an image, comprising:
presetting a reference object, and acquiring true color information of the reference object; capturing an image with the reference object as a training captured image with an imaging device;
detecting environmental spectrum information when the training shot image is shot as training environmental spectrum information through a spectrum probe balanced with the center of the optical axis of the imaging equipment;
receiving the training shot image, the training environment spectrum information and the true color information of the reference object to form a training set; analyzing the training shot image to obtain the pixel gray value of the reference object in the training shot image;
training according to the pixel gray value of the reference object in the training set and the true color information of the reference object to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the shot image under each environment spectrum information condition;
shooting a target image by using the imaging equipment, and detecting real-time environment spectrum information when the target image is shot by using the spectrum probe balanced with the center of the optical axis of the imaging equipment;
receiving the target image and the real-time environment spectrum information which are shot currently by the imaging equipment; analyzing the currently shot target image to obtain the gray value of each pixel of the target image;
obtaining a pixel gray compensation value of the target image according to the corresponding relation between the pixel gray value of the target image and the true color information of the image under the current environment information;
and obtaining a restored image of the currently shot target image according to the pixel gray compensation value of the target image and the gray value of each pixel of the target image.
2. The method for adaptively performing true color restoration on an image according to claim 1, further comprising:
in the training set, obtaining a pixel gray compensation value restored image of the reference object according to the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the environmental spectrum information and the pixel gray value of the reference object;
obtaining a loss value of image restoration according to the restored image of the reference object and the true color information of the reference object; when the loss value of the image restoration is smaller than or equal to a preset loss value threshold, storing the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectrum information condition;
and when the loss value of the image restoration is greater than the preset loss value threshold, generating an image of the restored image of the reference object and the loss function of the true color information.
3. The method for adaptively performing true color restoration on an image according to claim 2, further comprising:
in the training set, receiving an instruction for processing data in the training set, and analyzing the instruction to obtain a data identifier of a training image indicating deletion;
searching and deleting the data of the training image indicating deletion according to the data identification of the training image indicating deletion;
and obtaining the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectrum information condition according to the residual data in the training set.
4. The method of claim 2, wherein the loss function of the restored image and true color information is:
wherein,
pi(x, y) represents the image gray value of the reconstructed restored image at the ith pixel point, tiAnd (x, y) represents the image gray value of the real true color image at the ith pixel point.
5. The method for adaptively performing true color restoration on an image according to claim 1, further comprising:
in the training set, receiving a newly added training shot image, newly added training environment spectrum information and updated true color information of a reference object in the newly added training shot image to update data of the training set;
and training by using the updated data in the training set to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the updated environmental spectrum information.
6. A system for adaptively performing true color restoration of an image, comprising: the system comprises an image training data collector, an image data training analyzer, a real-time image data receiver, a real-time image compensation analyzer and a real-time image restorer; wherein,
the image training data collector is connected with the image data training analyzer and used for receiving training shot images of imaging equipment, training environment spectrum information of a spectrum probe balanced with the center of an optical axis of the imaging equipment and true color information of a reference object preset in the training shot images to form a training set; analyzing the training shot image to obtain the pixel gray value of the reference object in the training shot image;
the image data training analyzer is connected with the image training data collector and the real-time image data receiver and is used for training according to the pixel gray value of the reference object in the training set and the true color information of the reference object to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the shot image under the condition of environmental spectrum information;
the real-time image data receiver is connected with the image data training analyzer and the real-time image compensation analyzer and is used for receiving a current shot image of the imaging equipment and current environment spectrum information of the spectrum probe balanced with the center of the optical axis of the imaging equipment; analyzing the current shot image to obtain the gray value of each pixel of the current shot image;
the real-time image compensation analyzer is connected with the real-time image data receiver and the real-time image restorer and is used for obtaining a pixel gray level compensation value of the currently shot target image according to the corresponding relation between the pixel gray level value of the shot target image under the current environment information and the true color information of the image;
and the real-time image restorer is connected with the real-time image compensation analyzer and is used for obtaining a restored image of the target image according to the pixel gray level compensation value of the target image and the gray level value of each pixel of the target image.
7. The system for adaptively performing true color restoration on an image according to claim 6, further comprising: an image data training evaluator coupled to the image data training analyzer for:
in the training set, obtaining a restored image of the reference object according to the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the environmental spectrum information and the pixel gray value of the reference object;
obtaining a loss value of image restoration according to the restored image of the reference object and the true color information of the reference object; when the loss value of the image restoration is smaller than or equal to a preset loss value threshold, storing the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectrum information condition;
and when the loss value of the image restoration is greater than the preset loss value threshold, generating an image of the restored image of the reference object and the loss function of the true color information.
8. The system for adaptively performing true color restoration on an image according to claim 7, further comprising: a training image data processor coupled to the image data training evaluator to:
receiving an instruction for processing the data in the training set, and analyzing the instruction to obtain a data identifier of a training image which indicates deletion;
searching and deleting the data of the training image indicating deletion according to the data identification of the training image indicating deletion;
and obtaining the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the environment spectrum information condition according to the residual data in the training set.
9. The system for adaptively performing true color restoration on an image according to claim 7, wherein the loss function between the restored image and true color information is:
wherein,
pi(x, y) represents the image gray value of the reconstructed restored image at the ith pixel point, tiAnd (x, y) represents the image gray value of the real true color image at the ith pixel point.
10. The system for adaptively performing true color restoration on an image according to claim 6, further comprising: a training image data updater connected to the image data training analyzer and configured to:
receiving a newly added training shot image, newly added training environment spectrum information and updated true color information of a reference object in the newly added training shot image to update data of the training set;
and training by using the updated data in the training set to obtain the corresponding relation between the pixel gray value of the shot image and the true color information of the image under the condition of the updated environmental spectrum information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811143926.0A CN109274945B (en) | 2018-09-29 | 2018-09-29 | Method and system for self-adaptively performing true color restoration on image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811143926.0A CN109274945B (en) | 2018-09-29 | 2018-09-29 | Method and system for self-adaptively performing true color restoration on image |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109274945A true CN109274945A (en) | 2019-01-25 |
| CN109274945B CN109274945B (en) | 2020-05-22 |
Family
ID=65199143
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811143926.0A Active CN109274945B (en) | 2018-09-29 | 2018-09-29 | Method and system for self-adaptively performing true color restoration on image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109274945B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110763154A (en) * | 2019-11-25 | 2020-02-07 | 宁波五维检测科技有限公司 | Large-field-of-view microscopic 3D (three-dimensional) morphology multi-channel measuring device and method |
| CN111025324A (en) * | 2020-01-06 | 2020-04-17 | 青梧桐有限责任公司 | Household pattern generating method based on distance measuring sensor |
| CN111553953A (en) * | 2020-04-23 | 2020-08-18 | 国科天成（北京）科技有限公司 | System and method for calibrating pseudo color of night vision device |
| CN112399160A (en) * | 2019-08-19 | 2021-02-23 | Oppo广东移动通信有限公司 | Color management method and apparatus, terminal and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090295947A1 (en) * | 2008-06-03 | 2009-12-03 | Olympus Corporation | Imaging device |
| CN104410850A (en) * | 2014-12-25 | 2015-03-11 | 武汉大学 | Colorful digital image chrominance correction method and system |
| CN106791731A (en) * | 2016-12-01 | 2017-05-31 | 深圳市唯特视科技有限公司 | A kind of scene RGB restoration methods of combining camera spectral sensitivity information |
| CN106840398A (en) * | 2017-01-12 | 2017-06-13 | 南京大学 | A kind of multispectral light-field imaging method |
-
2018
- 2018-09-29 CN CN201811143926.0A patent/CN109274945B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090295947A1 (en) * | 2008-06-03 | 2009-12-03 | Olympus Corporation | Imaging device |
| CN104410850A (en) * | 2014-12-25 | 2015-03-11 | 武汉大学 | Colorful digital image chrominance correction method and system |
| CN106791731A (en) * | 2016-12-01 | 2017-05-31 | 深圳市唯特视科技有限公司 | A kind of scene RGB restoration methods of combining camera spectral sensitivity information |
| CN106840398A (en) * | 2017-01-12 | 2017-06-13 | 南京大学 | A kind of multispectral light-field imaging method |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112399160A (en) * | 2019-08-19 | 2021-02-23 | Oppo广东移动通信有限公司 | Color management method and apparatus, terminal and storage medium |
| CN110763154A (en) * | 2019-11-25 | 2020-02-07 | 宁波五维检测科技有限公司 | Large-field-of-view microscopic 3D (three-dimensional) morphology multi-channel measuring device and method |
| CN111025324A (en) * | 2020-01-06 | 2020-04-17 | 青梧桐有限责任公司 | Household pattern generating method based on distance measuring sensor |
| CN111553953A (en) * | 2020-04-23 | 2020-08-18 | 国科天成（北京）科技有限公司 | System and method for calibrating pseudo color of night vision device |
| CN111553953B (en) * | 2020-04-23 | 2020-12-11 | 国科天成（北京）科技有限公司 | System and method for calibrating pseudo color of night vision device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109274945B (en) | 2020-05-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109274945B (en) | Method and system for self-adaptively performing true color restoration on image | |
| CN105868689B (en) | A kind of face occlusion detection method based on concatenated convolutional neural network | |
| CN111967393A (en) | Helmet wearing detection method based on improved YOLOv4 | |
| CN103839085B (en) | A kind of detection method of compartment exception crowd density | |
| CN109711318B (en) | Multi-face detection and tracking method based on video stream | |
| CN107948465A (en) | A kind of method and apparatus for detecting camera and being disturbed | |
| Seckiner et al. | Forensic image analysis–CCTV distortion and artefacts | |
| CN110222596B (en) | Driver behavior analysis anti-cheating method based on vision | |
| CN107220633A (en) | A kind of intelligent mobile enforcement system and method | |
| CN110866512B (en) | Monitoring camera shielding detection method based on video classification | |
| CN112329719B (en) | Behavior recognition method, behavior recognition device and computer-readable storage medium | |
| US20220122360A1 (en) | Identification of suspicious individuals during night in public areas using a video brightening network system | |
| CN106339657A (en) | Straw incineration monitoring method and device based on monitoring video | |
| CN115546742A (en) | Rail foreign matter identification method and system based on monocular thermal infrared camera | |
| KR20200059643A (en) | ATM security system based on image analyses and the method thereof | |
| CN110780356A (en) | Subway platform clearance foreign matter detecting system | |
| KR102540208B1 (en) | Fire detection method using deep learning | |
| CN112132048A (en) | Community patrol analysis method and system based on computer vision | |
| AU2021240327A1 (en) | Infringement detection method, device and system | |
| Beghdadi et al. | Towards the design of smart video-surveillance system | |
| CN116229347A (en) | Crowd safety abnormal event identification method | |
| CN110516538B (en) | Prison double off-duty violation assessment method based on deep learning target detection | |
| CN112183287A (en) | People counting method of mobile robot under complex background | |
| Suresh et al. | An efficient low cost background subtraction method to extract foreground object during human tracking | |
| US11126852B2 (en) | Method for training a person recognition model using images from a camera, and method for recognizing persons from a trained person recognition model by means of a second camera in a camera network |
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 |