CN105787525A - Image detection method and image detection device - Google Patents
Image detection method and image detection device Download PDFInfo
- Publication number
- CN105787525A CN105787525A CN201410830944.1A CN201410830944A CN105787525A CN 105787525 A CN105787525 A CN 105787525A CN 201410830944 A CN201410830944 A CN 201410830944A CN 105787525 A CN105787525 A CN 105787525A
- Authority
- CN
- China
- Prior art keywords
- image
- sliding window
- decision tree
- variate
- split vertexes
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses an image detection method and an image detection device. By using variable selection characteristics of random forests, after decision trees are trained, determining can be carried out by automatically searching corresponding image characteristic variables on every split node of every decision tree. During the image detection using the above mentioned method, for every sliding window, the characteristic variable corresponding to every split node of every trained decision tree can be calculated, and then the random forest is used to identify whether the image in the sliding windows comprises the target object. After all the sliding windows are scanned, the image detection result can be acquired. By adopting the image detection method, not all the characteristic variables in the sliding windows are required to be calculated, only the characteristic variables corresponding to the split nodes of the decision trees are required to be calculated, and the number of the characteristic variables corresponding to the split nodes are way smaller than the number of all the characteristic variables in the sliding window, and therefore the image detection speed can be greatly improved by adopting the method provided by the invention.
Description
Technical field
It relates to technical field of image processing, particularly relate to a kind of image detecting method and device.
Background technology
Image detecting method is exactly detect destination object from given image, for instance, method for detecting human face, it is simply that detection comprises the part of face from given image.
For Face datection, utilize the method that SVM (SupportVectorMachine, support vector machine) grader and sliding window combine, sliding window scanning is carried out under different scale, to each sliding window, it is necessary to calculate 162338 haar eigenvalues (that is, characteristic variable).Then, utilize the SVM classifier trained judges whether comprise face in image.
As shown in the above, this kind of detection method needs to be accomplished by calculating whole characteristic variables of each sliding window before carrying out image recognition, and the scanning speed of such sliding window is very slow, namely image detection speed is very slow.
Summary of the invention
For overcoming Problems existing in correlation technique, the disclosure provides a kind of image detecting method and device, and technical scheme is as follows:
First aspect according to disclosure embodiment, it is provided that a kind of image detecting method, including:
For each sliding window of picture to be detected, obtain the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that training in advance obtains one by one；
The characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and criterion corresponding to described split vertexes, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
In conjunction with first aspect, in the first possible implementation of first aspect, the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that the described training in advance of acquisition one by one obtains, including:
For each described sliding window, set up the corresponding relation between the split vertexes of described decision tree and the computing formula of characteristics of image variable；
Inquire about described corresponding relation and determine the computing formula of described characteristics of image variable, and calculate the variate-value obtaining described characteristics of image variable according to described computing formula.
In conjunction with first aspect, in the implementation that the second of first aspect is possible, the described characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and criterion corresponding to described split vertexes, detect whether the image in described sliding window comprises destination object, including:
Whether characteristics of image variate-value that split vertexes according to described decision tree is corresponding and criterion corresponding to described split vertexes, comprise destination object to the image in described sliding window and vote；
According to each decision tree voting results to the image in described sliding window in random forest, it is determined that whether the image in described sliding window comprises destination object.
In conjunction with the implementation that the second of first aspect is possible, in the third possible implementation of first aspect, described according to each decision tree voting results to the image in described sliding window in random forest, it is determined that whether the image in described sliding window comprises destination object, including:
Add up the voting results of each decision tree in described random forest；
Whether results maximum for poll in all voting results is comprised as the image determined in described sliding window the judged result of destination object.
In conjunction with first aspect, in the 4th kind of possible implementation of first aspect, described method also includes:
Calculate the characteristics of image variate-value of sample image；
According to described characteristics of image variate-value, training obtains each decision tree in random forest.
Second aspect according to disclosure embodiment, it is provided that a kind of image detection device, including:
Acquisition module, for each sliding window for picture to be detected, obtains the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that training in advance obtains one by one；
Detection module, for the criterion that the characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and described split vertexes are corresponding, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
In conjunction with second aspect, in the first possible implementation of second aspect, described acquisition module includes:
Set up submodule, for for each described sliding window, setting up the corresponding relation between the split vertexes of described decision tree and the computing formula of characteristics of image variable；
First calculating sub module, determines the computing formula of described characteristics of image variable for inquiring about described corresponding relation, and calculates the variate-value obtaining described characteristics of image variable according to described computing formula.
In conjunction with second aspect, in the implementation that the second of second aspect is possible, described detection module includes:
Choose submodule in a vote, for the criterion that the characteristics of image variate-value corresponding according to the split vertexes of described decision tree and described split vertexes are corresponding, whether the image in described sliding window is comprised destination object and votes；
First determines submodule, for according to each decision tree voting results to the image in described sliding window in random forest, it is determined that whether the image in described sliding window comprises destination object.
In conjunction with the implementation that the second of second aspect is possible, in the third possible implementation of second aspect, described first determines that submodule includes:
Statistics submodule, for adding up the voting results of each decision tree in described random forest；
Second determines submodule, for whether results maximum for poll in all voting results comprises the judged result of destination object as the image determined in described sliding window.
In conjunction with second aspect, in the 4th kind of possible implementation of second aspect, described device also includes:
Second calculating sub module, for calculating the characteristics of image variate-value of sample image；
Training submodule, for according to described characteristics of image variate-value, training obtains each decision tree in random forest.
The third aspect according to disclosure embodiment, it is provided that a kind of terminal unit, including: processor；For storing the memorizer of processor executable；Wherein, described processor is configured to:
For each sliding window of picture to be detected, obtain the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that training in advance obtains one by one；
The characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and criterion corresponding to described split vertexes, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
Embodiment of the disclosure that the technical scheme of offer can include following beneficial effect: described image detecting method utilizes the variable selection characteristic of random forest, after the decision tree trained, the split vertexes place of every decision tree can be automatically found corresponding characteristics of image variable and judge.When utilizing the method to carry out image detection, for each sliding window, it is only necessary to calculate the characteristic variable corresponding to each split vertexes of each good decision tree of training in advance.Then, utilize whether the image in sliding window described in random forest identification comprises destination object.After scanning through whole sliding window, obtain image testing result.As shown in the above, this image detecting method need not calculate the whole characteristic variables in sliding window, only need to calculate the split vertexes characteristic of correspondence variable of decision tree, and the quantity of whole characteristic variables that the quantity of the characteristic variable corresponding to split vertexes is far smaller than in sliding window, therefore, utilize the method to carry out image detection and detection speed can be greatly improved.
It should be appreciated that above general description and details hereinafter describe and be merely illustrative of, the disclosure can not be limited.
Accompanying drawing explanation
Accompanying drawing herein is merged in description and constitutes the part of this specification, it is shown that meets embodiments of the invention, and is used for explaining principles of the invention together with description.
Fig. 1 is the flow chart of a kind of image detecting method according to an exemplary embodiment；
Fig. 2 is the flow chart of the step S110 in a kind of Fig. 1 according to an exemplary embodiment；
Fig. 3 is the flow chart of the step S120 in a kind of Fig. 1 according to an exemplary embodiment；
Fig. 4 is the flow chart of the another kind of image detecting method according to an exemplary embodiment；
Fig. 5 is a kind of image detection device block diagram according to an exemplary embodiment；
Fig. 6 is the block diagram of a kind of acquisition module 110 according to an exemplary embodiment；
Fig. 7 is the block diagram of a kind of detection module 120 according to an exemplary embodiment；
Fig. 8 is the block diagram that a kind of first according to an exemplary embodiment determines submodule 320；
Fig. 9 is the block diagram of the another kind of image detection device according to an exemplary embodiment；
Figure 10 is the block diagram of a kind of device 800 according to an exemplary embodiment；
Figure 11 is the block diagram of a kind of device 1900 according to an exemplary embodiment.
By above-mentioned accompanying drawing, it has been shown that the embodiment that the disclosure is clear and definite, will there is more detailed description hereinafter.These accompanying drawings are not intended to be limited the scope of disclosure design by any mode, but are the concept that those skilled in the art illustrate the disclosure by reference specific embodiment.
Detailed description of the invention
Before exemplary embodiment is illustrated, first introduce the concept of random forest.Random forest is to set up a forest by random manner, has a lot of decision trees inside forest, and namely random forest is a grader comprising multiple decision tree.After obtaining random forest, in time having the input of new sample, each decision tree in random forest is first allowed respectively this sample to be judged, it is determined which kind of this sample should belong to, and then, the ballot of each decision tree obtains the classification results of sample.
In detail below exemplary embodiment being illustrated, its example representation is in the accompanying drawings.When as explained below relates to accompanying drawing, unless otherwise indicated, the same numbers in different accompanying drawings represents same or analogous key element.Embodiment described in following exemplary embodiment does not represent all embodiments consistent with the disclosure.On the contrary, they only with in appended claims describe in detail, the disclosure some in the example of consistent apparatus and method.
Fig. 1 is the flow chart of a kind of image detecting method according to an exemplary embodiment, and the method can apply in mobile terminal (such as, smart mobile phone) or server.As it is shown in figure 1, the method may comprise steps of:
In step s 110, for each sliding window of picture to be detected, obtain the characteristics of image variate-value that in the random forest that training in advance obtains, the split vertexes of each decision tree is corresponding one by one.
Random forest has variable selection characteristic, namely trains each decision tree obtained, and can be automatically found corresponding variable at each split vertexes place and judge.So, when utilizing random forest to carry out image detection, as long as calculating the variate-value of characteristics of image variable corresponding with each split vertexes of decision tree in picture to be detected, it is not necessary to calculate the variate-value of all images characteristic variable in picture to be detected.
For Face datection, for each sliding window, it is necessary to the image in sliding window is first transformed into the picture of 24*24, the picture of each 24*24 has 162338 haar features, namely haar characteristic vector has 162338 variablees, and the variable that random forest is used is sub-fraction therein.Therefore, after training random forest, the tree construction of each decision tree in random forest also just secures, criterion that namely split vertexes in decision tree is corresponding it has been determined that and, each split vertexes which variable corresponding it has been determined that.For haar feature, 162338 variablees corresponding to the picture of each 24*24 have certain order, and each variable has the computational methods of correspondence.
In the step s 120, the characteristics of image variate-value that split vertexes according to each decision tree is corresponding, and the criterion that described split vertexes is corresponding, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
Utilize the random forest that training obtains that image is identified, namely utilize each decision tree in random forest that the classification of image is voted, maximum for poll are classified as the classification of this image.Whether the variate-value that decision tree is the characteristics of image variable according to each split vertexes meets corresponding criterion, and the classification of image is voted.Such as, when whether comprising face in detection image, if majority decision tree thinks that this image does not comprise face, then the final detection result of random forest is that this image does not comprise face.
Assume whether detection image comprises face, then destination object is exactly face；Assume whether comprise automotive license plate in detection image, then destination object is exactly automotive license plate.
Often one sliding window of scanning, whether the image continuing to identify in this sliding window comprises destination object, until the whole of picture to be detected scan through, obtains the testing result of whole picture to be detected.
The image detecting method that the present embodiment provides, utilizes the variable selection characteristic of random forest, and so, after the decision tree trained, the split vertexes place of every decision tree can be automatically found corresponding characteristics of image variable and judge.Utilize the method, when carrying out image detection, for each sliding window, it is only necessary to calculate the characteristic variable corresponding to each split vertexes of each good decision tree of training in advance.Then, utilize whether the image in sliding window described in random forest identification comprises destination object.After scanning through whole sliding window, obtain image testing result.As shown in the above, this image detecting method, the whole characteristic variables in sliding window need not be calculated before carrying out image detection, only need to calculate the split vertexes characteristic of correspondence variable of decision tree, and the quantity of whole characteristic variables that the quantity of the characteristic variable corresponding to split vertexes is far smaller than in sliding window, therefore, utilize the method to carry out image detection and detection speed can be greatly improved.
Fig. 2 is the method flow diagram of a kind of step S110 according to an exemplary embodiment, as in figure 2 it is shown, above-mentioned steps S110 can be realized by following steps:
In step S111, for each described sliding window, set up the corresponding relation between the split vertexes of each decision tree with the computing formula of corresponding characteristics of image variable in described random forest.
The computing formula of each characteristics of image variable is fixing, such as, for haar feature, the computing formula of the 1000th characteristics of image variable is: in current sliding window the third line, the vector of the 5th row is to the vector sum of the tenth row feature, deducts in current sliding window fourth line the vector of the 5th row to the vector sum of the tenth row.
In step S112, determine the computing formula of described characteristics of image variable according to described corresponding relation, and calculate the variate-value obtaining characteristics of image variable corresponding to described split vertexes according to described computing formula.
Obtaining the computing formula of characteristics of image variable corresponding to split vertexes according to the inquiry of described corresponding relation, the computing formula obtained according to inquiry calculates the variate-value of characteristics of image variable corresponding to this split vertexes.Such as, the variate-value obtaining this characteristics of image variable is calculated according to the computing formula of the 1000th characteristics of image variable, in order to the variate-value that later use split vertexes is corresponding carries out choosing in a vote of decision tree.
Corresponding relation between the computing formula of the characteristics of image variable that split vertexes that every decision tree needing to set up each sliding window is corresponding is corresponding with this split vertexes, so, when utilizing random forest to carry out image detection, could automatically calculate the variate-value of characteristics of image variable corresponding with the split vertexes of decision tree in picture to be detected, consequently facilitating picture to be detected is detected by random forest.
Fig. 3 is the method flow diagram of the step S120 according to an exemplary embodiment, as it is shown on figure 3, step S120 can be realized by following steps:
In step S121, the characteristics of image variate-value corresponding according to the split vertexes of described decision tree and criterion corresponding to described split vertexes, whether the image in described sliding window is comprised destination object and votes.
The testing result of sliding window, according to criterion corresponding to the variate-value of characteristics of image variable corresponding to each split vertexes and this split vertexes, is voted by each decision tree.
In step S122, according to each decision tree voting results to the image in described sliding window in random forest, it is determined that whether the image in described sliding window comprises destination object.
For any one sliding window in picture to be detected, the image in sliding window is classified by each decision tree according to the tree construction of self.In statistics random forest, the voting results of each decision tree predict the classification results of this sliding window, if it is considered to the voting results that the image in this sliding window comprises destination object occupy the majority, then random forest predicts that the image of this sliding window comprises destination object；If it is considered to the voting results that the image of this sliding window does not comprise destination object occupy the majority, then random forest predicts that the image of this sliding window does not comprise destination object.
Random forest predicts the testing result of picture to be detected according to the voting results of each decision tree, and, picture to be detected is carried out the sliding window scanning of different scale, often one sliding window of scanning, detects whether this sliding window comprises destination object.Scan through whole sliding window and obtain the testing result of whole picture to be detected.Such as, during the face in detection image, it is possible to detect the whole faces comprised in picture to be detected.
Fig. 4 is the flow chart of the another kind of image detecting method according to an exemplary embodiment, and the method is applied in mobile terminal or server.Before utilizing random forest to carry out image detection, it is necessary to utilize substantial amounts of sample image to train each decision tree obtaining in random forest.
As shown in Figure 4, the method can also comprise the following steps on the basis of Fig. 1:
In step S410, calculate all images characteristic variable value of sample image.
Sample image is zoomed to the size of template image, then, calculates all images characteristic variable value in the image after convergent-divergent.The number of whole characteristics of image variable of the template image of such as 24*24 is 162338.
In the step s 420, the described characteristics of image variate-value according to sample image, training obtains each decision tree in random forest.
Utilizing all images characteristic variable value of sample image, training decision tree structure obtains the criterion of the split vertexes of decision tree, and training obtains each decision tree composition random forest.Then perform step S110, utilize random forest that picture to be detected is detected.
The image detecting method that the present embodiment provides, before utilizing random forest to carry out image detection, need to utilize the training of substantial amounts of sample image to obtain random forest, namely the tree construction of each decision tree in random forest is obtained, then, utilize the random forest that training obtains detects in image whether comprise destination object.
Fig. 5 is the block diagram of a kind of image detection device according to an exemplary embodiment, and this device is applied in mobile terminal (such as, smart mobile phone) or server.As it is shown in figure 5, this device includes acquisition module 110 and detection module 120；
Acquisition module 110 is configured to, and for each sliding window of picture to be detected, obtains the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that training in advance obtains one by one.
Detection module 120 is configured to, the characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and criterion corresponding to described split vertexes, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
The image detection device that the present embodiment provides, utilizes the variable selection characteristic of random forest, and after the decision tree trained, the split vertexes place of every decision tree can be automatically found corresponding characteristics of image variable and judge.When utilizing the method to carry out image detection, for each sliding window, it is only necessary to calculate the characteristic variable corresponding to each split vertexes of each good decision tree of training in advance.Then, utilize whether the image in sliding window described in random forest identification comprises destination object.After scanning through whole sliding window, obtain image testing result.As shown in the above, this image detection device need not calculate the whole characteristic variables in sliding window, only need to calculate the split vertexes characteristic of correspondence variable of decision tree, and the quantity of whole characteristic variables that the quantity of the characteristic variable corresponding to split vertexes is far smaller than in sliding window, therefore, this device can be greatly improved detection speed.
Fig. 6 is the block diagram of a kind of acquisition module 110 according to an exemplary embodiment, and as shown in Figure 6, described acquisition module 110 includes setting up submodule 210 and the first calculating sub module 220；
Set up submodule 210 to be configured to, for each described sliding window, set up the corresponding relation between the split vertexes of described decision tree and the computing formula of characteristics of image variable.
The computing formula of each characteristics of image variable is fixing, such as, for haar feature, the computing formula of the 1000th characteristics of image variable is: in current sliding window the third line, the vector of the 5th row is to the vector sum of the tenth row feature, deducts in current sliding window fourth line the vector of the 5th row to the vectorial sum of the tenth row.
First calculating sub module 220 is configured to, and inquires about described corresponding relation and determines the computing formula of described characteristics of image variable, and calculates the variate-value obtaining described characteristics of image variable according to described computing formula.
Obtaining the computing formula of characteristics of image variable corresponding to split vertexes according to the inquiry of described corresponding relation, the computing formula obtained according to inquiry calculates the variate-value of characteristics of image variable corresponding to this split vertexes.
Fig. 7 is the block diagram of a kind of detection module 120 according to an exemplary embodiment, as it is shown in fig. 7, described detection module 120 can include choosing submodule 310 and first in a vote determines submodule 320；
Choose submodule 310 in a vote to be configured to, the characteristics of image variate-value corresponding according to the split vertexes of described decision tree and criterion corresponding to described split vertexes, whether the image in described sliding window is comprised destination object and votes.
First determines that submodule 320 is configured to, according to each decision tree voting results to the image in described sliding window in random forest, it is determined that whether the image in described sliding window comprises destination object.
Fig. 8 is the block diagram that first according to an exemplary embodiment determines submodule 320, and described first determines that submodule 320 can include statistics submodule 410 and second and determine submodule 420.
Statistics submodule 410 is configured to, and adds up the voting results of each decision tree in described random forest.
Second determines that submodule 420 is configured to, and whether results maximum for poll in all voting results comprises the judged result of destination object as the image determined in described sliding window.
If it is considered to the voting results that the image in this sliding window comprises destination object occupy the majority, then random forest predicts that the image of this sliding window comprises destination object；If it is considered to the voting results that the image of this sliding window does not comprise destination object occupy the majority, then random forest predicts that the image of this sliding window does not comprise destination object.
Random forest predicts the testing result of picture to be detected according to the voting results of each decision tree, and, picture to be detected is carried out the sliding window scanning of different scale, often one sliding window of scanning, detects in this sliding window whether comprise destination object.Scan through whole sliding window and obtain the testing result of whole picture to be detected.Such as, during the face in detection image, it is possible to detect the whole faces comprised in picture to be detected.
Fig. 9 is the block diagram of the another kind of image detection device according to an exemplary embodiment, as it is shown in figure 9, described device also includes computing module 510 and training module 520 on the basis of embodiment illustrated in fig. 5.
Computing module 510 is configured to, and calculates the characteristics of image variate-value of sample image.
Training module 520 is configured to, and according to described characteristics of image variate-value, training obtains each decision tree in random forest.
The image detection device that the present embodiment provides, before utilizing random forest to carry out image detection, need to utilize the training of substantial amounts of sample image to obtain random forest, namely the tree construction of each decision tree in random forest is obtained, then, utilize the random forest that training obtains detects in image whether comprise destination object.
About the device in above-described embodiment, the concrete mode that wherein modules performs to operate has been described in detail in about the embodiment of the method, and explanation will be not set forth in detail herein.
Figure 10 is the block diagram of a kind of device 800 for realizing image detecting method according to an exemplary embodiment.Such as, device 800 can be mobile phone, computer, digital broadcast terminal, messaging devices, game console, tablet device, armarium, body-building equipment, personal digital assistant etc..
As shown in Figure 10, device 800 can include following one or more assembly: processes assembly 802, memorizer 804, power supply module 806, multimedia groupware 808, audio-frequency assembly 810, the interface 812 of input/output (I/O), sensor cluster 814, and communications component 816.
Process assembly 802 and generally control the integrated operation of device 800, such as with display, call, data communication, the operation that camera operation and record operation are associated.Process assembly 802 and can include one or more processor 820 to perform instruction, to complete all or part of step of above-mentioned method.Additionally, process assembly 802 can include one or more module, it is simple to what process between assembly 802 and other assemblies is mutual.Such as, process assembly 802 and can include multi-media module, with facilitate multimedia groupware 808 and process between assembly 802 mutual.
Memorizer 804 is configured to store various types of data to support the operation at device 800.The example of these data includes any application program for operation on device 800 or the instruction of method, contact data, telephone book data, message, picture, video etc..Memorizer 804 can be realized by any kind of volatibility or non-volatile memory device or their combination, such as static RAM (SRAM), Electrically Erasable Read Only Memory (EEPROM), Erasable Programmable Read Only Memory EPROM (EPROM), programmable read only memory (PROM), read only memory (ROM), magnetic memory, flash memory, disk or CD.
The various assemblies that power supply module 806 is device 800 provide electric power.Power supply module 806 can include power-supply management system, one or more power supplys, and other generate, manage and distribute, with for device 800, the assembly that electric power is associated.
Multimedia groupware 808 includes the screen providing an output interface between described device 800 and user.In certain embodiments, screen can include liquid crystal display (LCD) and touch panel (TP).If screen includes touch panel, screen may be implemented as touch screen, to receive the input signal from user.Touch panel includes one or more touch sensor to sense the gesture on touch, slip and touch panel.Described touch sensor can not only sense the border of touch or sliding action, but also detects the persistent period relevant to described touch or slide and pressure.In certain embodiments, multimedia groupware 808 includes a front-facing camera and/or post-positioned pick-up head.When device 800 is in operator scheme, during such as screening-mode or video mode, front-facing camera and/or post-positioned pick-up head can receive the multi-medium data of outside.Each front-facing camera and post-positioned pick-up head can be a fixing optical lens system or have focal length and optical zoom ability.
Audio-frequency assembly 810 is configured to output and/or input audio signal.Such as, audio-frequency assembly 810 includes a mike (MIC), and when device 800 is in operator scheme, during such as call model, logging mode and speech recognition mode, mike is configured to receive external audio signal.The audio signal received can be further stored at memorizer 804 or send via communications component 816.In certain embodiments, audio-frequency assembly 810 also includes a speaker, is used for exporting audio signal.
I/O interface 812 provides interface for processing between assembly 802 and peripheral interface module, above-mentioned peripheral interface module can be keyboard, puts striking wheel, button etc..These buttons may include but be not limited to: home button, volume button, startup button and locking press button.
Sensor cluster 814 includes one or more sensor, for providing the state estimation of various aspects for device 800.Such as, what sensor cluster 814 can detect device 800 opens/closed mode, the relative localization of assembly, such as described assembly is display and the keypad of device 800, the position change of all right detecting device 800 of sensor cluster 814 or 800 1 assemblies of device, the presence or absence that user contacts with device 800, the variations in temperature of device 800 orientation or acceleration/deceleration and device 800.Sensor cluster 814 can include proximity transducer, is configured to when not having any physical contact object near detection.Sensor cluster 814 can also include optical sensor, such as CMOS or ccd image sensor, for using in imaging applications.In certain embodiments, this sensor cluster 814 can also include acceleration transducer, gyro sensor, Magnetic Sensor, pressure transducer or temperature sensor.
Communications component 816 is configured to facilitate between device 800 and other equipment the communication of wired or wireless mode.Device 800 can access the wireless network based on communication standard, such as WiFi, 2G or 3G, or their combination.In one exemplary embodiment, communications component 816 receives the broadcast singal or the broadcast related information that manage system from external broadcasting via broadcast channel.In one exemplary embodiment, described communications component 816 also includes near-field communication (NFC) module, to promote junction service.Such as, can based on RF identification (RFID) technology in NFC module, Infrared Data Association (IrDA) technology, ultra broadband (UWB) technology, bluetooth (BT) technology and other technologies realize.
In the exemplary embodiment, device 800 can be realized by one or more application specific integrated circuits (ASIC), digital signal processor (DSP), digital signal processing appts (DSPD), PLD (PLD), field programmable gate array (FPGA), controller, microcontroller, microprocessor or other electronic components, is used for performing said method.
In the exemplary embodiment, additionally providing a kind of non-transitory computer-readable recording medium including instruction, for instance include the memorizer 804 of instruction, above-mentioned instruction can have been performed said method by the processor 820 of device 800.Such as, described non-transitory computer-readable recording medium can be ROM, random access memory (RAM), CD-ROM, tape, floppy disk and optical data storage devices etc..
A kind of non-transitory computer-readable recording medium, when the instruction in described storage medium is performed by the processor of mobile terminal so that terminal unit is able to carry out a kind of image detecting method, and described method includes:
For each sliding window of picture to be detected, obtain the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that training in advance obtains one by one；
The characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and criterion corresponding to described split vertexes, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
Figure 11 is the block diagram of a kind of device 1900 for realizing image detecting method according to an exemplary embodiment.Such as, device 1900 may be provided in a server.As shown in figure 11, device 1900 includes processing assembly 1922, and it farther includes one or more processor and the memory resource representated by memorizer 1932, can by the instruction of the execution processing assembly 1922 for storing, for instance application program.In memorizer 1932 application program of storage can include one or more each corresponding to the module of one group of instruction.It is configured to perform instruction, the embodiment of the method corresponding to perform above-mentioned Fig. 1～Fig. 4 additionally, process assembly 1922.
Device 1900 can also include a power supply module 1926 and be configured to perform the power management of device 1900, and a wired or wireless network interface 1950 is configured to be connected to device 1900 network and input and output (I/O) interface 1958.Device 1900 can operate based on the operating system being stored in memorizer 1932, for instance WindowsServerTM, MacOSXTM, UnixTM, LinuxTM, FreeBSDTM or similar.
Those skilled in the art, after considering description and putting into practice invention disclosed herein, will readily occur to other embodiment of the present invention.The application is intended to any modification of the present invention, purposes or adaptations, and these modification, purposes or adaptations are followed the general principle of the present invention and include the undocumented known general knowledge in the art of the disclosure or conventional techniques means.Description and embodiments is considered only as exemplary, and the true scope of the present invention and spirit are pointed out by claim below.
It should be appreciated that the invention is not limited in precision architecture described above and illustrated in the accompanying drawings, and various amendment and change can carried out without departing from the scope.The scope of the present invention is only limited by appended claim.
Claims (11)
1. an image detecting method, it is characterised in that including:
For each sliding window of picture to be detected, obtain the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that training in advance obtains one by one；
The characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and criterion corresponding to described split vertexes, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
2. method according to claim 1, it is characterised in that the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that the described training in advance of acquisition one by one obtains, including:
For each described sliding window, set up the corresponding relation between the split vertexes of described decision tree and the computing formula of characteristics of image variable；
Inquire about described corresponding relation and determine the computing formula of described characteristics of image variable, and calculate the variate-value obtaining described characteristics of image variable according to described computing formula.
3. method according to claim 1, it is characterized in that, the described characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and criterion corresponding to described split vertexes, detect whether the image in described sliding window comprises destination object, including:
Whether characteristics of image variate-value that split vertexes according to described decision tree is corresponding and criterion corresponding to described split vertexes, comprise destination object to the image in described sliding window and vote；
According to each decision tree voting results to the image in described sliding window in random forest, it is determined that whether the image in described sliding window comprises destination object.
4. method according to claim 3, it is characterised in that described according to each decision tree voting results to the image in described sliding window in random forest, it is determined that whether the image in described sliding window comprises destination object, including:
Add up the voting results of each decision tree in described random forest；
Whether results maximum for poll in all voting results is comprised as the image determined in described sliding window the judged result of destination object.
5. method according to claim 1, it is characterised in that described method also includes:
Calculate the characteristics of image variate-value of sample image；
According to described characteristics of image variate-value, training obtains each decision tree in random forest.
6. an image detection device, it is characterised in that including:
Acquisition module, for each sliding window for picture to be detected, obtains the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that training in advance obtains one by one；
Detection module, for the criterion that the characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and described split vertexes are corresponding, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
7. device according to claim 6, it is characterised in that described acquisition module includes:
Set up submodule, for for each described sliding window, setting up the corresponding relation between the split vertexes of described decision tree and the computing formula of characteristics of image variable；
First calculating sub module, determines the computing formula of described characteristics of image variable for inquiring about described corresponding relation, and calculates the variate-value obtaining described characteristics of image variable according to described computing formula.
8. device according to claim 6, it is characterised in that described detection module includes:
Choose submodule in a vote, for the criterion that the characteristics of image variate-value corresponding according to the split vertexes of described decision tree and described split vertexes are corresponding, whether the image in described sliding window is comprised destination object and votes；
First determines submodule, for according to each decision tree voting results to the image in described sliding window in random forest, it is determined that whether the image in described sliding window comprises destination object.
9. device according to claim 8, it is characterised in that described first determines that submodule includes:
Statistics submodule, for adding up the voting results of each decision tree in described random forest；
Second determines submodule, for whether results maximum for poll in all voting results comprises the judged result of destination object as the image determined in described sliding window.
10. device according to claim 6, it is characterised in that described device also includes:
Second calculating sub module, for calculating the characteristics of image variate-value of sample image；
Training submodule, for according to described characteristics of image variate-value, training obtains each decision tree in random forest.
11. a terminal unit, it is characterised in that including:
Processor；
For storing the memorizer of processor executable；
Wherein, described processor is configured to:
For each sliding window of picture to be detected, obtain the characteristics of image variate-value corresponding to split vertexes of each decision tree in the random forest that training in advance obtains one by one；
The characteristics of image variate-value corresponding according to the split vertexes of decision tree each described and criterion corresponding to described split vertexes, detect whether the image in described sliding window comprises destination object, until scanning through whole sliding windows of described picture to be detected, obtain image testing result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410830944.1A CN105787525A (en) | 2014-12-26 | 2014-12-26 | Image detection method and image detection device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410830944.1A CN105787525A (en) | 2014-12-26 | 2014-12-26 | Image detection method and image detection device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105787525A true CN105787525A (en) | 2016-07-20 |
Family
ID=56389019
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410830944.1A Pending CN105787525A (en) | 2014-12-26 | 2014-12-26 | Image detection method and image detection device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105787525A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650637A (en) * | 2016-12-02 | 2017-05-10 | 华中师范大学 | Smiling face detector based on condition random forests and method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103679132A (en) * | 2013-07-15 | 2014-03-26 | 北京工业大学 | A sensitive image identification method and a system |
| CN103902591A (en) * | 2012-12-27 | 2014-07-02 | 中国科学院深圳先进技术研究院 | Decision tree classifier establishing method and device |
-
2014
- 2014-12-26 CN CN201410830944.1A patent/CN105787525A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103902591A (en) * | 2012-12-27 | 2014-07-02 | 中国科学院深圳先进技术研究院 | Decision tree classifier establishing method and device |
| CN103679132A (en) * | 2013-07-15 | 2014-03-26 | 北京工业大学 | A sensitive image identification method and a system |
Non-Patent Citations (2)
| Title |
|---|
| 詹曙等: "基于随机森林的脑磁共振图像分类", 《电子测量与仪器学报》 * |
| 郭天楚等: "基于改进的随机森林的人体部件识别", 《中国传媒大学学报自然科学版》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106650637A (en) * | 2016-12-02 | 2017-05-10 | 华中师范大学 | Smiling face detector based on condition random forests and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105279898A (en) | Alarm method and device | |
| CN105512685A (en) | Object identification method and apparatus | |
| CN105095881A (en) | Method, apparatus and terminal for face identification | |
| CN104933170A (en) | Information exhibition method and device | |
| CN105828201A (en) | Video processing method and device | |
| CN106202194A (en) | The storage method and device of screenshot picture | |
| CN103902689A (en) | Clustering method, incremental clustering method and related device | |
| CN104899610A (en) | Picture classification method and device | |
| CN104700353A (en) | Image filter generating method and device | |
| CN105354560A (en) | Fingerprint identification method and device | |
| CN106021092A (en) | Method and device for determining starting duration of application program | |
| CN104503888A (en) | Warning method and device | |
| CN105631803A (en) | Method and device for filter processing | |
| CN104063865A (en) | Classification model creation method, image segmentation method and related device | |
| CN105205494A (en) | Similar picture identification method and device | |
| CN104537380A (en) | Clustering method and device | |
| CN105139378A (en) | Card boundary detection method and apparatus | |
| CN106096009A (en) | Method for generating message and device | |
| CN104077597A (en) | Image classifying method and device | |
| CN104484683A (en) | Porn picture detection method and device | |
| CN104035673A (en) | Object control method and relevant device | |
| CN105808647A (en) | Message editing processing method and device | |
| CN105188027A (en) | Nearby user searching method and device | |
| CN105260743A (en) | Pattern classification method and device | |
| CN104268149A (en) | Clustering method and clustering device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20160720 |