Summary of the invention
In view of this, the invention reside in provides a kind of face identification method with prompt facility, and to solve in the above-mentioned existing identifying, employing voice suggestion user adjusts decipherment distance and aligns degree, has the slower problem of recognition speed.
For addressing the above problem, the invention provides a kind of face identification method with prompt facility and comprise:
In the identifying, show the facial image that camera is caught at screen, present simultaneously the organ characteristic's of corresponding people's face prompting figure at described screen; Wherein, described prompting figure is used for showing organ characteristic on the described facial image and the applying degree of described prompting figure, and described applying degree is used to indicate distance between user and the camera and the degree that aligns of user and camera.
Further, the figure of prompting shown in is comprised of orthogonal prompting framework or prompting line segment; Described prompting framework or prompting line segment comprise horizontal and vertical two parts; Wherein, described horizontal prompting framework or prompting line segment are used for showing the applying degree with people's eye feature, and described framework or the prompting line segment of pointing out longitudinally is used for showing applying degree with people's nose feature at least.
Further, described prompting framework or prompting line segment are T shape or cruciform.
Further, the figure of prompting shown in is the prompting framework;
Wherein, the size of described prompting framework when being effective decipherment distance between people's face and the camera, the shared area of pixel of people's eye, nose feature on the described facial image that shows on the screen.
Further, the position of described prompting figure on described screen fixed;
Or, detect the organ characteristic of the people's face on the described facial image of every frame, on described every two field picture, organ characteristic's first reference point of people's face of two described detections of correspondence is set at least;
On described prompting figure, the second identical with described the first reference point quantity, that the position is corresponding reference point is set;
When the described every frame facial image that shows at described screen, described prompting figure, described the first reference point, the second reference point are coincided.
Further, when showing described prompting figure, the organ characteristic of the described people's face of output prompting and the information of described prompting figure applying degree;
Wherein, described information is used for the prompting user adjustment with respect to the position and attitude of described camera, and the organ characteristic of described people's face and described prompting figure are fitted.
Further, described information adopts character or figure to be presented on the described screen, or adopts sound to play.
Further, when showing described prompting figure, also comprise:
When screen shows every two field picture, detect the facial image in this two field picture, the T shape framework that presents the pixel that comprises the feature of eye, nose in the described facial image, the two ends of the horizontal framework in this T shape framework respectively with facial image in two tail of the eyes of human eye coincide, the area of vertical framework of this T shape framework is for comprising at least the area of the pixel of the feature of nose in the facial image;
Or, when screen shows every two field picture, detect the facial image in this two field picture, present T shape line segment, the two ends of the horizontal line segment of this T shape line segment respectively with facial image in the pixel of two pupils of eye coincide, the nose feature towards facial image of vertical line segment of this T shape line segment is extended.
Further, also comprise: in the organ characteristic's of the described people's of presenting face prompting figure, present the use help information of described prompting figure.
Further, described prompting figure, the color of following figure are arbitrary transparency greater than zero and are different from the color of described image on screen color.
The user can check the organ characteristic of the people's face that shows in the prompting figure on the screen, by self move the organ characteristic that adjusts the people's face on the screen and prompting figure align degree and with the process of camera distance in, identification equipment can constantly detect facial image and identification.When the organ characteristic of people's face acquires a certain degree with the identical or applying of pointing out figure, just can effectively identify; Can also reach the best identified effect when coincideing or fit when reaching optimum degree, not only can effectively identifying.
Embodiment
For clearly demonstrating the scheme among the present invention, the below provides preferred embodiment and is described with reference to the accompanying drawings.
Embodiment one
Referring to Fig. 1, may further comprise the steps:
S11: after identification equipment starts, show the facial image that camera is caught at screen;
S12: present the prompting figure at described screen;
Wherein, described prompting figure is used for showing that the organ characteristic of the people's face on the described facial image and the applying degree of described prompting figure, described applying degree are used to indicate distance between user and the camera and the degree that aligns of user and camera.
This prompting figure can be prompting framework, prompting line segment, also can be the picture with organ characteristic's shape of people's face.The figure that prompting framework or prompting line segment can adopt the organ characteristic with people's face to match, T shape such as corresponding people's face eye feature, nose, mouth feature, it also can be cruciform, it also can be the picture of organ characteristic's shape of people's face, such as the picture of eye shape, the picture of nose shape, the picture of mouth shape etc., has certain position relationship between these eyes, nose, the mouth picture.
By these prompting figures, can show the organ characteristic of the people's face on the facial image and coincideing or the applying degree of these figures, with the distance between prompting user and the camera.For example: the face image of finding oneself as the user is less in the prompting framework, can move forward, with the distance between adjustment and the camera; And adjust the degree that aligns of self and camera, eye feature is aligned and take in the horizontal prompting framework, at least the pixel that comprises the nose feature in vertical prompting framework on the screen, the nose feature aligns and takes in vertical prompting framework, or nose feature and mouth feature align and take in vertical prompting framework, aligns and take framework until adjust to the organ characteristic of user's face.These features can be contour features, such as the profile of eyes, the profile of nose, the profile of mouth etc.
The prompting figure is other color that is different from the screen, the color of prompting figure can for arbitrary transparency greater than zero and be different from the color of described image on screen.
The user can check the organ characteristic of the people's face that shows in the prompting figure on the screen, by self move the organ characteristic that adjusts the people's face on the screen and prompting figure align degree and with the process of camera distance in, identification equipment can constantly detect facial image and identification.When the organ characteristic of people's face acquires a certain degree with the identical or applying of pointing out figure, just can effectively identify; Can also reach the highest recognition accuracy when coincideing or fit when reaching top, not only can effectively identifying.
The prompting figure can be different form, describes the detection effect of various prompting figures in detail below by embodiment.
Embodiment two
In embodiment two, the prompting figure adopts the form of prompting framework, referring to Fig. 2, may further comprise the steps:
S21: after identification equipment starts, show the facial image that camera is caught at screen;
Referring to Fig. 3, the facial image of display capture on the screen.
S22: the prompting figure that presents the organ characteristic of people's face on the described screen;
In the present embodiment, the prompting figure is the prompting framework of T shape, and on screen the fixed position at center.
The area of prompting framework is when being effective decipherment distance between people's face and the camera, and the area of the eye of described facial image, nose feature can also be the area that comprises the mouth feature.
The horizontal prompting frame body of T shape framework is for the applying degree of demonstration and people's eye feature, and vertically the prompting frame body is used for demonstration and people's nose feature and the applying degree of mouth feature for applying degree or vertical prompting frame body of demonstration and people's nose feature.
S23: identification equipment detects the distance of user and camera and aligns degree, and sends suggestion voice.
Referring to Fig. 4, owing to being used for distant apart from camera, the organ characteristic of people's face is less in the prompting framework, the identification equipment None-identified.At this moment, the user can be before the camera the mobile or voice suggestion of " please move forward " of sending according to identification equipment, adjust and camera between distance, the schematic diagram after the adjustment can be referring to Fig. 5.
Sometimes the user does not align camera, when screen imaging, situation as shown in Figure 6 can occur, and namely user's head has certain deflection, and the face feature on screen does not all show.Be presented as that user's eye feature might exceed the horizontal framework of T shape prompting framework, the nose feature has exceeded vertical framework of T shape prompting framework.Owing to the deflection of user's head, vertically go up the straight line of nose feature, mouth Characteristics creation, can be at a certain angle with vertical line shape, nose feature and/or mouth feature have exceeded vertical framework of T shape prompting framework.
At this moment the user can according to self face-image and screen on the prompting framework the degree that aligns, or " please left deflection head " that send according to identification equipment, " please's to the right deflection head " voice suggestion, adjust the degree that aligns facial and camera, make face align camera, so that facial characteristics is in the prompting framework, screen show as shown in Figure 7 to plus effect.
In this embodiment, the user is according to the distance of pointing out framework adjustment self and camera and align degree, because on screen, user's face-image in the prompting frame that shows and the deviation between the prompting frame, distinguish out from visually being very easy to, the user can slightly adjust the distance of self and camera and align degree according to the situation of screen display.
Preferably, except aligning the degree of employing voice suggestion user and camera, can also adopt the degree that aligns of character or graphical cues user and camera.Display reminding user " please left deflection head ", " please's to the right deflection head " literal for example.
The size of prompting framework can be a large amount of by statistics, the shared area of the organ characteristic of people's face on the screen when identify successfully, are optimized according to the area that counts, and the area of framework is pointed out in acquisition.
For example can determine in the following manner the framework size, through large quantitative statistics facial image, determine people's interpupillary distance between 50~70mm, preferred 60mm.According to be between user and the camera discrimination the highest apart from the time, determine the length of the pixels across point that 60mm shows at screen, and in the centre of the length of pixels across point, extend downward vertically the vertical pixel isometric with the length of pixels across point, thereby determine T shape line, distance with 40 pixel length of distance T shape line forms the T shape framework of surrounding this T shape line, and the area of this T shape framework is exactly the area of prompting framework.
Therefore, when the organ characteristic of user's face in the prompting framework, and when substantially taking framework, identification equipment just can effectively identify user's identity, take the organ characteristic of framework, the feature that may comprise whole eyes, nose, mouth also may only have eye feature and a part of nose and/or mouth feature.Voice suggestion compared to existing technology, because the user can clearly judge by the prompting figure and camera between distance and align degree, distance between rapid adjustment and the camera, align degree, be convenient to identification equipment and obtain best recognition image, thereby fast identification, shortened recognition time, recognition efficiency is high.
In an embodiment of the present invention, except the framework that adopts T shape, can also adopt the cruciform framework, or adopt the as shown in Figure 8 form of a horizontal framework, a vertical framework combination, or adopt the as shown in Figure 9 form of two eye feature frameworks, a nose feature framework combination, and the position between two the eye feature frameworks, nose feature framework is fixed.
Except the prompting framework, can also adopt the form of prompting line segment, comprise and laterally point out as shown in figure 10 line segment, vertically point out line segment combination, or T shape is pointed out line segment as shown in figure 11, or the form of the prompting line segment combination of prompting line segment, a nose feature of two corresponding eyes as shown in figure 12, and the position between two the eye features prompting line segments, nose feature prompting line segment is fixed.When adopting T shape line, when the distance between user and the camera be discrimination the highest apart from the time, the two ends of T shape line are close with two pupils on the facial image or overlap on the screen.
When adopting the prompting line segment, the user can be according to the organ characteristic of people's face and coincideing or the applying degree of prompting line segment, adjusts the distance of self and camera or align degree.The user can be by vertical line segment adjustment of T shape line and the degree that aligns of camera, the process that vertical framework adjustment by prompting frame among the process of adjusting and Fig. 6 aligns degree is identical, and needs longitudinally indication wire section are positioned on the symmetrical center line of facial image and get final product.
The prompting figure can also adopt prompting framework or the prompting line segment of the eye that comprises the people, nose picture shape, and as shown in figure 13, the shape of the prompting line segment of employing is more pressed close to people's eye, nose shape, is easilier understood by the user.
In the above-described embodiments, preferably,, except aligning the degree of employing voice suggestion user and camera, can also adopt the degree that aligns of graphical cues user and camera.For example adopt the form of following figure.
In step S22, when showing fixedly prompting frame, referring to Figure 14, on the feature of the people's face that detects, present the figure of following of or close shape identical with described fixedly prompting frame.Be T shape framework or T shape line segment etc. such as following figure; With the position between prompting user adjustment and the camera, by following figure the feature of face and prompting figure are fitted, or be contained in the prompting figure.
Follow the distance dependent between figure and user and the camera, during close together, the shape of following figure is larger; When distant, the shape of following figure is less.
During specific implementation, can be by the facial image that detects in every two field picture, on the pixel of at least eye of facial image, nose, mouth feature, present the color of following figure.
The color of following figure is identical with the color of prompting figure, also can adopt two kinds of different colors, but follows the overlap degree of figure reminding user between the head portrait that screen shows and the prompting figure of fixing.When following figure with the prompting graphs coincide, the distance between user and the camera is the highest distance of discrimination, is also referred to as the best identified distance, and the degree that aligns between user and the camera is in the highest degree that aligns, and at this moment, recognition effect is best.
Preferably, when the prompting figure was prompting frame, following figure can be prompting frame or prompting line segment; Follow figure and be certain proportion relation, the Range-based between this proportionate relationship and people's face and the camera with the prompting figure.
When the display reminding figure, can be when screen shows every two field picture, present and follow figure, when screen shows every two field picture, detect the facial image in this two field picture, the T shape framework that presents the pixel that comprises the feature of eye, nose in the described facial image, the two ends of the horizontal framework in this T shape framework respectively with facial image in two tail of the eyes of human eye coincide, the area of vertical framework of this T shape framework is for comprising at least the area of the pixel of the feature of nose in the facial image;
Or, when screen shows every two field picture, detect the facial image in this two field picture, present T shape line segment, the two ends of the horizontal line segment of this T shape line segment respectively with facial image in the pixel of two pupils of eye coincide, the nose feature towards facial image of vertical line segment of this T shape line segment is extended.
In this embodiment, prompting framework or the position of prompting line segment in screen are fixed, and certainly, prompting framework or prompting line segment can also be movable, and the facial characteristics of following the user.Below by embodiment three explanations.
Embodiment three
Referring to Figure 15, in this embodiment, may further comprise the steps:
S31: after identification equipment starts, show the facial image that camera is caught at screen;
Show the facial image as shown in Figure 3 that camera is caught on the screen.
S32: the position of organ characteristic on screen of constantly detecting the people's face on every frame facial image;
S33: when showing every two field picture, at the organ characteristic's of people's face position display prompting figure;
Can pass through following dual mode display reminding figure;
First kind of way:
Detect the organ characteristic's of the people's face on the image the position of pixel, show the prompting figure of the position that surrounds this pixel.The position of the position of the organ characteristic's of people's face pixel in the prompting figure can according to circumstances be set.For example set at random or specify in a particular location in the prompting figure.
The second way:
1) organ characteristic of the people's face on every two field picture partly arranges the first reference point of the organ characteristic of at least two corresponding people's faces;
Can select every two field picture by the detection method of existing facial image, the organ characteristic of the people's face on the detected image, such as eye feature etc., on the image after the binaryzation, identify the position, two canthus of eye, with two pixels of this position as the first reference point.
2) on described prompting figure, the second identical with described the first reference point quantity, that the position is corresponding reference point is set;
On described prompting figure, if what adopt is the prompting framework, can on the position on the horizontal close both sides of framework, second reference point be set respectively respectively; If the prompting line segment also can laterally respectively arrange respectively second reference point on the close position on both sides of prompting line segment; Distance between two the second reference points equals two distances between the first reference point, and is positioned at horizontal framework or laterally points out the centre of line segment.
When 3) showing every frame facial image, prompting figure on the described screen, described the first reference point, the second reference point are coincided.
The facial image that Figure 16 shows is positioned at the upper right corner of screen, after the user moves, demonstrates the facial image that is positioned at the lower left corner of screen shown in Figure 17 at screen.By above-mentioned dual mode, although the user is moving with respect to camera, when showing facial image on the screen, the organ characteristic of people's face all can fit with the prompting figure.
S34: identification equipment detects the distance of user and camera and aligns degree, and sends suggestion voice.
The user is in moving process, by detecting the change in location of facial image in the consecutive frame image, to realize following of prompting figure and people's face section.The user can be according to the degree of agreement of the facial image that shows on the screen with the prompting framework, judge the degree that aligns of the distance of self and camera and self and camera, adjust the position relationship between self and the camera, thereby realize that the user in moving process, realizes identifying fast user's identity.
By the step of this embodiment, the user can be in moving process, adjusts the distance between self and the camera and aligns degree.Because the facial characteristics that the prompting figure is followed the people, with respect to fixing prompting figure, user's adjustment process has improved user's experience when having simplified identification.
Preferably, in the above-described embodiment, the ratio of the ratio between the distance between user and the camera and the effective decipherment distance, the degree that aligns when aligning degree and effectively identifying can be by the formal output of screen with numeral or figure or sound.When showing, can respectively with the digital form demonstration of number percent, can also show ratio by the form of colored scroll bar in the fixed frame.
Preferably, in the above-described embodiments, in the time of the display reminding frame, the use help information of output prompting frame can be shown to the user at screen by the form of literal, also can pass through the formal output of sound to the user.
In the above-described embodiments, the large I of prompting figure arranges according to the mode of mentioning among the embodiment two, also can for example carry out relative set in conjunction with parameters such as camera, Processing Algorithm, screen sizes: the image that camera is taken is 320 * 240 pixels or the larger image of pixel quantity, can from image, detect the organ characteristic's of people's face pixel, for the larger image of pixel quantity, can remove some pixels from image, keep the pixel of respective numbers according to recognizer.
For example: the camera (imaging size is 640*480) that adopts 30W, when the decipherment distance that is between user and the camera farthest, the pixel of the people's face that detects is less, by selecting 81 precalculated positions on the facial image that detects, belonging to the organ characteristic's of people's face
Vegetarian refreshments, with each pixel in these 81 pixels as the center, forming a plurality of different n*n is matrix, obtain the zone of a plurality of n*n pixel, for example, 3*3,4*4,6*6 etc., calculate these regional eigenwerts, such as gray scale etc., with the eigenwert that calculates and in advance the respective regions of the image of registration eigenwert relatively, whether judge is same user's recognition result.The final comparative result that obtains 81 pixel positions according to these comparative results, is judged the facial image of current identification whether identical with the facial image of in advance registration.
When being in different decipherment distances between user and the camera, the position relationship between the prompting figure of user fixed position on image that screen shows and screen is also different, and the effect of identification is also different.
Referring to Figure 18, the imaging size of camera is 640*480, far away when the user distance camera, stand in the C place, position of decipherment distance farthest, at this moment the facial image of camera collection screen show less, display effect as shown in Figure 4, the facial image that detects from the image that gathers only has 200*200 pixel, when identifying by recognizer, because pixel is less, can exist identification inaccurate, the problem that recognition effect is poor.
Nearer when the user distance camera, stand in the A place, position of decipherment distance farthest, at this moment the facial image of camera collection screen show larger, the facial image that detects from the image that gathers has 480*480 pixel, although it is more to detect the pixel of people's face, because the user distance camera is nearer, can there be distortion in the image of camera edge, exist pixel corresponding to the image of distortion when identify by recognizer, can cause the problem that identification is inaccurate, mistake is identified.
Be the B place, position of best decipherment distance when the distance of user distance camera, at this moment the facial image of camera collection goodness of fit with the prompting figure on screen is the highest, such as Fig. 5, Fig. 7, Fig. 9, Figure 10 etc., the facial image that detects from the image that gathers has 300*300 pixel.
Referring to Figure 19, the imaging size of camera is 640*480, if adopt the prompting line segment of T shape at screen prompting figure, the prompting line segment comprises 151*151 pixel, adopt the guiding of prompting figure, distance between user and the camera is that the highest distance of discrimination and facial image are when the highest with prompting figure applying degree on screen, the pixel of two pupil positions of the eyes feature of people's face approaches or overlaps the horizontal two ends of T shape line on this moment screen, and the central spot of the horizontal prompting line segment of T shape line is in the centre of people's face left-right symmetric center line.At this moment, discrimination is the highest, and is the most accurate.
If the prompting frame of T shape then comprises 250*250 pixel in the prompting frame.Large like this prompting frame, just in time corresponding to be between user and the camera discrimination the highest apart from the time, the pixel at the position, canthus of eyes is near the horizontal both sides of T shape prompting frame, or just in time overlaps.At this moment, discrimination is the highest, and is the most accurate.
The pixel quantity that prompting figure in the above embodiments comprises not is certain, screen for different size, by the prompting figure that the various difform positions of a large amount of tests are fixed, size is constant, the user be in best identified apart from the time, all can determine the quantity of pixel of the organ characteristic's who comprises people's face prompting figure.
For the method for setting forth among each embodiment of the present invention, within the spirit and principles in the present invention all, any modification of doing, be equal to replacement, improvement etc., all should be included in guarantor of the present invention
Within the scope.