CN102968624A

CN102968624A - Method for positioning human eyes in human face image

Info

Publication number: CN102968624A
Application number: CN2012105350089A
Authority: CN
Inventors: 修春波
Original assignee: Tianjin Polytechnic University
Current assignee: Tianjin Polytechnic University
Priority date: 2012-12-12
Filing date: 2012-12-12
Publication date: 2013-03-13

Abstract

The invention belongs to the machine vision and image processing field and specifically relates to a method for positioning human eyes in a human face image, wherein a polar-coordinate gray-level integral projection method is employed to perform human eye positioning. Polar coordinates are established with the central position of the human face as the pole, the horizontally leftward ray of the pole as the pole axis and the anti-clockwise direction as the positive direction of the angle. The gray level of the image is projected in the direction of the polar angle so as to determine the angles of the eyes; then, the pixels in the direction of the eyes are subjected to gray-level integral projection in the horizontal direction and the position of the minimal projection value is the projection position of the human eyes; therefore, the coordinate positions of the human eyes are determined and the human eye detection requirement of the inclined human face image is satisfied. The method for positioning the human eyes in the human face image can be applied to the human eye identification field.

Description

Human-eye positioning method in a kind of facial image

Technical field

The invention belongs to machine vision and image processing field, relate to the human-eye positioning method in a kind of facial image, particularly a kind of method of under polar coordinates, carrying out integral projection realization human eye location.

Background technology

How the main research in human eye location detects human eye in given facial image and locates, be the hot research problem of machine vision and image processing field, also be one of key technical problem of the researchs such as driver tired driving monitoring, eyeball mouse, eye tracking always.At present more relatively ripe algorithm is being arranged aspect the human eye location, as the Edge Gradient Feature method, based on the Hough conversion detection method, template matching method, geometric properties method, based on the detection method of AdaBoost sorter, gray-level projection method etc.Although these methods are relatively ripe, each own different advantages and disadvantages.Existing gray-level projection method is people's face to be carried out the projection of horizontal and vertical directions, human eye is located in distributed intelligence according to Wave crest and wave trough, this method realizes simply, locating speed is very fast, but the attitude to human eye has stricter requirement, when people's face plane rotates when causing human eye to have certain angle of inclination, be difficult to carry out effective location according to the distribution of Wave crest and wave trough.

Therefore, a kind of integral projection human-eye positioning method that can adapt in people's face rotation situation of design has certain using value.

Summary of the invention

Technical matters to be solved by this invention is, designs the human-eye positioning method in a kind of facial image, is particularly useful for the requirement of the human eye location in people's face rotation situation.

The technical solution adopted in the present invention is: the human-eye positioning method in a kind of facial image, adopt the method for polar coordinates gray-level projection to carry out the human eye location.Take people's face center as limit, take limit level ray left as pole axis, get counter clockwise direction for the positive dirction of angle, set up polar coordinates with this.The gray scale of image is carried out projection along the polar angle direction, to determine the angle at eyes place, again the pixel of eyes place direction is carried out gray-level projection on horizontal direction, the position of projection value minimum is the projected position of human eye, determine thus the coordinate position of human eye, and the human eye detection requirement of satisfied inclination facial image.

The object of the invention is to utilize polar gray-level projection information to determine the position of human eye in people's face, to adapt to the requirement of human eye location in the rotation people face.

Description of drawings

Fig. 1 is human eye position fixing process figure.

Fig. 2 is the integral projection figure of polar angle direction.

Fig. 3 is the integral projection filtering figure of polar angle direction.

Fig. 4 is left eye direction pixel gray-level projection figure in the horizontal direction.

Fig. 5 is right eye direction pixel gray-level projection figure in the horizontal direction.

Embodiment

Below in conjunction with embodiment and accompanying drawing the present invention is described in further detail.

Fig. 1 (a) is depicted as with human face region.Because always there are some isolated noises in the facial image that obtains under the natural conditions, in order to eliminate the impact of noise on human eye location, can adopt median filter method to remove the impact of noise.Adopt 3 * 3 median filter method that original image is carried out filtering among the present invention.Adopt large Tianjin method (maximum variance between clusters) definite threshold to carry out binary conversion treatment to filtered image.Image after the binary conversion treatment is shown in Fig. 1 (b).

Take people's face center as limit, take limit level ray left as pole axis, get counter clockwise direction for the positive dirction of angle, set up polar coordinates with this.

To the image after the binaryzation, in [0 °, 360 °] polar angle scope, carry out gray-level projection by formula (1).

Wherein, I (ρ, θ) is θ for polar angle, and the utmost point directly is the gray-scale value of the pixel of ρ.Polar angle gray-level projection figure for accurately determining the pole location of integral projection, adopts the mean filter method that its filtering is processed as shown in Figure 2, and filtered result as shown in Figure 3.In the filtered perspective view, three corresponding people's of trough district difference eyes and the angle of mouth.Because the corresponding angular regions of mouth is wider, therefore the corresponding minimal point of the widest trough can be defined as the mouth direction, be the human eye direction with the orientation determination of remaining two minimal points.As shown in Figure 3, the scope of three troughs is (0 °, 88 °), (88 °, 175 °) and (175 °, 360 °), can judge that therefore (0 °, 88 °) is for comprising the zone of left eye, (88 °, 175 °) for comprising the zone of right eye, (175 °, 360 °) are for comprising the zone of mouth.The angle of getting the gray-level projection minimum value that comprises left eye region is the left eye angle, the angle that comprises the gray-level projection minimum value of right eye region is the right eye angle, acquired results is shown in Fig. 1 (c), and left eye place angle is 50 °, and right eye place angle is 150 °.

With the pixel of left eye place angle in the gray level image to the horizontal direction projection, the gained projection result as shown in Figure 4 because eyes are black, so the minimum point position is the projected position of left eye in the projection result.With the pixel of right eye place angle in the gray level image to the horizontal direction projection, the gained projection result as shown in Figure 5, the minimum point position is the projected position of right eye in the projection result.Thereby can try to achieve the corresponding utmost point electrical path length of left eye and right eye, the utmost point of left eye directly is 15, and the utmost point of right eye directly is 16.The polar coordinates position that can determine thus left eye is (15,50 °), and the polar coordinates position of right eye is (16,150 °), and the position of right and left eyes is shown in the cross curve among Fig. 1 (d).

The invention has the advantages that, under polar coordinates, utilize the gray-level projection of facial image to determine position of human eye, still can carry out effective location to human eye for the people's face that rotates.

Claims

1. the human-eye positioning method in the facial image is characterized in that, adopts the method for polar coordinates gray-level projection to carry out the human eye location.

2. the human-eye positioning method in a kind of facial image according to claim 1 is characterized in that, take people's face center as limit, take limit level ray left as pole axis, gets counter clockwise direction for the positive dirction of angle, sets up polar coordinates with this.

3. according to claim 1, the human-eye positioning method in 2 described a kind of facial images, it is characterized in that, the gray scale of image is carried out projection along the polar angle direction, to determine the angle at eyes place, again the pixel of eyes place direction is carried out gray-level projection on horizontal direction, the position of projection value minimum is the projected position of human eye, determines thus the coordinate position of human eye.