Background technology
Video monitoring is widely used in safety and supervision area, is that public safety provides strong number with personnel's supervision
According to support.But relatively low towards the intelligent analysis degree also ratio of monitor video at present, leave in the human body towards monitor video
Context of detection, there is no disclosed in be specifically designed for the technological means of this application.Human body leaves detection and is mainly used in monitoring special scenes
Middle personnel have or not and leave designated area, have important function in field of intelligent video surveillance.Human body towards monitor video leaves
Detection is the important application of monitor video image procossing.Its handling process is:Obtain view data first from monitor video, so
Initialize detection zone afterwards and extract human body complexion, analyze color characteristic in detection zone further, and then realize human body and leave
Detection.
Leave the links in detection for human body, existing method such as patent 201510488408.2 adopts frame difference method
Moving region in detection video, if human body transfixion, can't detect;Patent 201310405276.3,
201310116469.7 extracting characteristics of human body's structural classification device to be trained, patent 201110264004.7 combines background subtraction
With human testing grader, patent 201010218630.8 is using the human body contour outline template detection multi-pose people with ambiguity
Body, patent 201310415544.X extracts, based on the colored human body detecting method with depth information, union feature, the feature obtaining
For human testing.Said method processing speed is very slowly it is impossible to be used for the common calculating platform such as DSP, ARM.Patent
201110026465.0 human testing is carried out based on depth image, it is not suitable for the monitor video image of routine.
Content of the invention
Leave a detection difficult problem for existing human body, on the basis of analysis of key link existing method deficiency, the present invention carries
Go out a kind of human body towards intelligent monitoring and leave reliable detection method, including the initialization of human testing region, based on colour of skin cluster
Human body complexion extract, based on color characteristic analysis human testing totally three part.The present invention is by the use of features of skin colors as human body
Detection, with reference to foundation, can reliably detect out human body target, have processing speed faster simultaneously.
Technical scheme in the present invention is described below:
1st, the human testing region based on Face datection and contouring head initializes
Judge whether human body leaves and must specify a reference area, if using whole video pictures as detection zone, nothing
Method distinguishes the multiple human body of same picture, and can increase amount of calculation, reduces processing speed.In the present invention, primary detection is arrived
Upper half of human body region as follow-up detection zone.As shown in Fig. 2 determining comprising the following steps that of this region:
Step1:Using the good Face datection grader based on deep learning of training in advance, in conjunction with contouring head template, detect
Go out human head location;
Step2:The head zone top left co-ordinate of hypothesis human body is (x1,y1), wide and high respectively w1、h1;The upper part of the body of human body
Region top left co-ordinate is (x2,y2), wide and high respectively w2、h2;Then according to data statistics rule, can substantially estimate human body
Upper part of the body position:
x2=x1-w1
y2=y1-h1/4
w2=w1×3
h2=h1×15/4
Innovative point is:
From video, the traditional algorithm processing speed in extracting directly human motion region is relatively slow and the degree of accuracy is not high, and some algorithms hold
Easily non-human moving target is judged to human body, some algorithms are difficult to static human body is detected.And according to Face datection joint head
Contouring, can detect static or motion human body target exactly, accurately delimit out human body and leave detection zone.
2nd, extracted based on the human body complexion of colour of skin cluster
Human body complexion has larger discrimination with clothing, the color of environment, and the colour of skin of different people also slightly difference.Special based on this
Point, can be according to human body skin tone testing human body target.After initialization human testing region, the colour of skin is used to gather in this region of this frame
Class, obtains one group of characteristic value based on hsv color space, for describing human body complexion information.As shown in figure 3, concrete steps are such as
Under:
Step1:Color cluster is carried out to the human testing region after initialization;
Step2:Skin color range is positioned according to head position, obtains Skin Color Information histogram feature;
Step3:Skin Color Information histogram feature is mapped to hsv color space, obtains the spy of Skin Color Information described in one group of collection
Value indicative.
Innovative point is:
Clustered using the colour of skin, extract the human body complexion feature based on hsv color model, accuracy in detection is high, processing speed is fast.
3rd, the human testing based on color characteristic analysis
Detection zone is left for human body, analyzes its color characteristic, mated with human body complexion feature.As shown in figure 4, it is concrete
Step is as follows:
Step1:For human testing region, mated with human body complexion feature pixel-by-pixel;
Step2:Statistics class colour of skin points, such as fruit colour of skin points are less than threshold value Th, then judge that human body leaves;If detection zone
Wide, high respectively w2, h2, according to data statistics rule, can value be:
Th=(w2×h2)/45
Innovative point is:
Analyze detection zone color characteristic based on the human body complexion extracting, effective detection can go out human body target, improve people
Body accuracy in detection.
Specific embodiment
With reference to diagram, the preferred embodiments of the present invention are described in detail.
It is as shown in Figure 1 that the human body of the present invention leaves detection workflow.Read a frame video image first;Then pass through pre-
The Face datection grader first training is processed with reference to contouring head template, if detecting and navigate to head, basis
Its position initialization human body leaves detection zone;Then for this extracted region features of skin colors;Finally divide in every frame afterwards
Analyse this region, verify whether it meets color condition;If be unsatisfactory for, judge that human body leaves.
Human testing region initialization flow process is as shown in Figure 2.First by the good Face datection grader of training in advance, tie
Syncephalon contouring template, detects human head location;Then according to data statistic analysis, you can human body is determined by head position
Leave detection zone.
It is as shown in Figure 3 that human body complexion extracts flow process.After initialization detection zone, first color is carried out to detection zone
Cluster;Then can determine the scope of colour of skin classification according to head position, and then ask for Skin Color Information histogram feature;Finally will
Histogram feature is mapped to hsv color space, obtains the characteristic value of Skin Color Information described in one group of collection.
Human testing flow process based on color characteristic analysis is as shown in Figure 4.In every frame first after initialization, by human body
Detection zone and human body complexion feature carry out Similarity matching, determine whether class colour of skin point;Then count the class skin of whole region
Color dot, obtains matching area, if less than threshold value, then judges that human body leaves.
It should be appreciated that for those of ordinary skills, can be improved according to the above description or be converted,
Such as change application etc., and all these modifications and variations all should belong to the protection domain of claims of the present invention.
Technical scheme in the embodiment of the present invention is clearly and completely described it is clear that described embodiment is this
Bright a part of embodiment, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not having
There is the every other embodiment being obtained under the premise of making creative work, broadly fall into the scope of protection of the invention.