Three-wavelength skin screening imaging system and method based on chopping technology
Technical Field
The invention relates to the technical field of skin identification, in particular to a three-wavelength skin screening imaging system and method based on a chopping technology.
Background
Modern biological feature recognition technology is more and more advanced, fingerprint recognition, face recognition, iris recognition, voiceprint recognition and other technologies are more and more mature, and the technology is gradually widely applied to production and life. Due to the convenience of face recognition, face recognition technology is rapidly developed in recent years, and particularly a face image processing technology based on an optical imaging technology is rapidly developed, and recognition algorithms of people on face features, facial bone features and the like are more mature, but the development of auxiliary identification technology based on living skin is slow. With the occurrence of the use of face masks and high-definition face images to fool recognition systems, it is a need to develop a technique for detecting whether a face image is collected from real skin.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a three-wavelength skin screening imaging system and method based on a chopping technology. The three filters of 450nm, 500nm and 550nm are arranged on the chopper, so that an imaging camera which is used for shooting through the chopper sequentially obtains image information in three spectral ranges, then whether the relation of the brightness change of each point along with time meets 1:0.4:0.9 is judged, whether the position of an imaging point is the skin position is determined, if so, the position is reserved, and if not, the position is deleted, thereby obtaining an image of the position of the real living skin only.
The aim of the invention is realized by the following technical scheme: a three-wavelength skin screening imaging system based on a chopping technology is characterized in that: the device comprises a chopping tri-spectral acquisition module, a chopping spectrum signal screening processing module and a chopping tri-spectral output module; the chopping spectrum signal screening and processing module is connected with the chopping spectrum output module.
Further defined, the tri-spectral acquisition module includes a chopper and an industrial camera.
Further defined, the chopper is configured with a first filter, a second filter, a third filter, a fourth filter, a fifth filter, and a sixth filter.
Further defined, the first filter, the second filter, the third filter, the fourth filter, the fifth filter and the sixth filter are respectively 450nm filter, 500nm filter, 550nm filter, 450nm filter, 500nm filter and 550nm filter, and the light intensities corresponding to the first filter, the second filter, the third filter, the fourth filter, the fifth filter and the sixth filter in sequence are A 450nm 、A 500nm 、A 550nm 、A 450nm 、A 500nm And A 550nm 。
Further limited, the chopper further comprises a rotation axis, the chopper rotates around the rotation axis clockwise or anticlockwise at a set frequency F when spectrum acquisition is carried out, so that the light rays are acquired and shot by an industrial camera with a frame frequency set as F after passing through the filter plate, and video image streams of F frames in total are acquired within 1 second, wherein the first frame of picture, the second frame of picture, … and the F frame of picture.
Further limiting, the chopped spectrum signal screening processing module screens any point P in the f-frame video image stream acquired by the industrial camera ij Screening and obtaining the corresponding point P 1ij 。
A three-wavelength skin screening imaging method based on a chopping technology specifically comprises the following steps:
the S0001 chopper rotates clockwise or anticlockwise around the rotation axis at a set frequency F, so that the light rays pass through the filter plate and are collected and shot by the industrial camera with the frame frequency set as F, and video image streams of F frames in total are collected within 1 second, wherein a first frame of picture, a second frame of picture, … and an F frame of picture are collected;
s0002 defines an arbitrary point P on the video image space ij Position: at any point P corresponding to the first frame picture 1ij Its light intensity is A 1ij Corresponding arbitrary point P in the second frame picture 2ij Its light intensity is A 2ij …, at arbitrary point P corresponding to the f-th frame picture fij Its light intensity is A 3ij ;
S0003 the light intensity of the pixel position of the skin changes with time for a period of T=1/(2F), wherein the light intensity corresponding to the 450nm filter is A 450nm The light intensity corresponding to the 500nm filter is A 500nm The light intensity corresponding to the 550nm filter is A 550nm Screening the images of any point;
S0004A obtained by screening ’ 1ij With an arbitrary point P on the video image space ij Position in the image corresponding to position P in the first frame picture 1ij Correspondingly, the light intensity A ’ 1ij Substitution ofIntensity A of the original position 1ij A first skin screening picture frame may be obtained, a second skin screening picture frame may be obtained, and so on, to form a processed video image stream.
Further defined, the screening is performed on the images of any point, and the screening conditions are as follows:
if satisfy A 450nm :A 500nm :A 550nm Approximately 1:0.4:0.9, then A ’ ij =A 450nm ;
If not satisfy A 450nm :A 500nm :A 550nm Approximately 1:0.4:0.9, then A ’ ij =0。
The beneficial effects of the invention are as follows: the invention utilizes three filters of 450nm, 500nm and 550nm to be arranged on the chopper, so that an imaging camera which is used for shooting through the chopper sequentially obtains image information in three spectral ranges, judges whether the relation of the brightness change of each point along with time meets 1:0.4:0.9, determines whether the imaging point position is the skin position, if so, the imaging point position is reserved, and if not, the imaging point position is deleted, thereby obtaining the image of the position where only the real living skin is located.
Drawings
FIG. 1 is a schematic diagram of a three-wavelength skin screening imaging technique based on a chopping technique;
in the figure, 1-chopper tri-spectral acquisition, 2-chopper tri-spectral signal screening processing, 3-chopper tri-spectral output, 4-face recognition, 11-chopper, 111-first filter, 112-third filter, 113-third filter, 114-fourth filter, 115-fifth filter, 116-sixth filter, 12-industrial camera, 13-video image stream, 131-first frame picture, 132-second frame picture, 133-third frame picture, 14-arbitrary point P on video image space ij 141-arbitrary point P 1ij 142-arbitrary point P 2ij 143-arbitrary point P fij Any point P in 21-video image ij Corresponding oscillogram, 22-skin spectrum screening flow, 31-processed video image flow, 32-processed face screening picture frame, 321-first face screening picture frame, 322-a second face screening picture frame;
FIG. 2 is a diagram of a face recognition system based on in vivo face skin detection;
in the figure, 1-chopper tri-spectral acquisition, 2-chopper tri-spectral signal screening processing, 3-chopper tri-spectral output, 4-face recognition, 11-chopper, 111-first filter, 112-third filter, 113-third filter, 114-fourth filter, 115-fifth filter, 116-sixth filter, 117-rotation axis, 12-industrial camera, 13-video image stream, 131-first frame picture, 132-second frame picture, 133-third frame picture, 14-arbitrary point P on video image space ij 141-arbitrary point P 1ij 142-arbitrary point P 2ij 143-arbitrary point P fij Any point P in 21-video image ij Corresponding oscillogram, 22-skin spectrum screening flow, 31-processed video image flow, 32-processed face screening picture frame, 321-first face screening picture frame, 322-second face screening picture frame, 41-image size and contour judgment, 42-face recognition algorithm, 43-recognition failed, 44-recognition passed.
FIG. 3 is a diagram of an in-car people counting system based on living face skin detection;
in the figure, 1-chopper tri-spectral acquisition, 2-chopper tri-spectral signal screening processing, 3-chopper tri-spectral output, 4-face recognition, 11-chopper, 111-first filter, 112-third filter, 113-third filter, 114-fourth filter, 115-fifth filter, 116-sixth filter, 117-rotation axis, 12-industrial camera, 13-video image stream, 131-first frame picture, 132-second frame picture, 133-third frame picture, 14-arbitrary point P on video image space ij 141-arbitrary point P 1ij 142-arbitrary point P 2ij 143-arbitrary point P fij Any point P in 21-video image ij Corresponding waveform diagram, 22-skin spectrum screening flow, 31-processed video image flow, 32-processed face screening picture frame, 321-first face screening picture frame, 322-second face screening picture frame, 32-face screening picture frame, 31-video image flow, 4-in-vehicle people statistics, 41-reserved area statistics count,face area and contour feature judgment 42, in-vehicle position judgment 43, screening, and 44-statistical output of the number of people.
Detailed Description
The technical solution of the present invention will be described in further detail with reference to the accompanying drawings, but the scope of the present invention is not limited to the following description.
As shown in fig. 1, a three-wavelength skin screening imaging system based on chopper technology is characterized in that: the device comprises a chopping tri-spectral acquisition module 1, a chopping spectrum signal screening processing module 2 and a chopping tri-spectral output module 3; the chopping spectrum signal screening and processing module 2 is connected with the chopping spectrum output module 3.
The three-spectrum acquisition module 1 comprises a chopper 11 and an industrial camera 12.
The chopper 11 is provided with a first filter 111, a second filter 112, a third filter 113, a fourth filter 114, a fifth filter 115, and a sixth filter 116.
The first filter 111, the second filter 112, the third filter 113, the fourth filter 114, the fifth filter 115 and the sixth filter 116 are respectively 450nm filter, 500nm filter, 550nm filter, 450nm filter, 500nm filter and 550nm filter, and the light intensities corresponding to the above filters in sequence are A 450nm 、A 500nm 、A 550nm 、A 450nm 、A 500nm And A 550nm 。
The chopper 11 further comprises a rotation axis 117, and the chopper rotates around the rotation axis 117 clockwise or anticlockwise at a set frequency F during spectrum acquisition, so that the light passes through the filter plate and is acquired and shot by the industrial camera 12 with the frame frequency set as F, and a video image stream 13 with F frames in total is acquired within 1 second, wherein a first frame picture 131, a second frame picture 132, … and an F frame picture 133 are acquired.
The chopped spectrum signal screening processing module 2 screens any point P in the f-frame video image stream 13 acquired by the industrial camera 12 ij 14 screening to obtain corresponding point P 1ij 。
A three-wavelength skin screening imaging method based on a chopping technology specifically comprises the following steps:
the S0001 chopper rotates around the rotation axis 117 clockwise or anticlockwise at a set frequency F, so that the light rays pass through the filter plate and are collected and shot by the industrial camera 12 with the frame frequency set as F, and a video image stream of F frames in total is collected within 1 second, wherein a first frame picture 131, a second frame picture 132, … and an F frame picture 133;
s0002 defines an arbitrary point P on the video image space ij Position: at the corresponding arbitrary point P in the first frame picture 131 1ij 141 with a light intensity A 1ij Corresponding arbitrary point P in the second frame picture 132 2ij 142 has a light intensity A 2ij …, at an arbitrary point P corresponding to the f-th frame picture 133 fij 143 with a light intensity A 3ij ;
S0003 the light intensity of the pixel position of the skin changes with time for a period of T=1/(2F), wherein the light intensity corresponding to the 450nm filter is A 450nm The light intensity corresponding to the 500nm filter is A 500nm The light intensity corresponding to the 550nm filter is A 550nm Screening the images of any point;
S0004A obtained by screening ’ 1ij With an arbitrary point P on the video image space ij 14 in the first frame picture 131 in the image corresponds to P 1ij Correspondingly, the light intensity A ’ 1ij Intensity A of light substituted for original position 1ij A first skin screening picture frame 321 may be obtained, a second skin screening picture frame 322 may be obtained, and so on to the resulting set of skin screening picture frames 32, thereby forming a processed video image stream 31.
Preferably, the screening is performed on the images of any point, and the screening conditions are as follows:
if satisfy A 450nm :A 500nm :A 550nm Approximately 1:0.4:0.9, then A ’ ij =A 450nm ;
If not satisfy A 450nm :A 500nm :A 550nm Approximately 1:0.4:0.9, then A ’ ij =0。
Embodiment 1:
as shown in fig. 2, a face recognition system based on living human face skin detection.
In the face recognition process, in order to prevent the occurrence of the behavior of fraudulently punching cards by using high-definition pictures when wearing the face mask, a face recognition system based on the skin detection of living human faces can be adopted. The system comprises four parts, namely a chopping tri-spectrum acquisition part 1, a chopping spectrum signal screening part 2, a chopping tri-spectrum output part 3 and a face recognition part 4.
The chopper tri-spectral acquisition 1 comprises a chopper 11 and an industrial camera 12. The chopper 11 is provided with a450nm first filter 111, a second 500nm filter 112, a third 550nm filter 113, a fourth 450nm filter 114, a500nm fifth filter 115, and a sixth 550nm filter 116.
The chopper 11 rotates at a frequency F, and the industrial camera 12 photographs the outside through the filter. The frame rate of the industrial camera 12 is f, so that within 1 second an f-frame video image stream 13 can be acquired, which is represented as a first frame picture 131, a second frame picture 132, …, an f-frame picture 133. Arbitrary point P on video image space ij 14 positions: at the corresponding arbitrary point P in the first frame picture 131 1ij 141 with a light intensity A 1ij Corresponding arbitrary point P in the second frame picture 132 2ij 142 has a light intensity A 2ij …, at an arbitrary point P corresponding to the f-th frame picture 133 fij 143 with a light intensity A 3ij . The video image stream 13 enters the chopped tri-spectral signal screening process 2 for processing.
Any point P in video image ij The corresponding waveform diagram 21 shows: the light intensity of the pixel position of the living human skin changes with time for a period T=1/(2F), wherein the light intensity corresponding to the 450nm filter is A 450nm The light intensity corresponding to the 500nm filter is A 500nm The light intensity corresponding to the 550nm filter is A 550nm The skin spectrum screening process 22 is used to screen images at any point: satisfy A 450nm :A 500nm :A 550nm Approximately 1:0.4: in the case of 0.9, A ’ ij =A 450nm The method comprises the steps of carrying out a first treatment on the surface of the Not satisfy A 450nm :A 500nm :A 550nm In the case of ≡ 1:0.4:0.9, A ’ ij =0. After screening treatment, the chopped tri-spectral output 3 part is processed.
A obtained by screening ’ 1ij With an arbitrary point P on the video image space ij 14 in the first frame picture 131 in the image corresponds to P 1ij Correspondingly, the light intensity A is used ’ 1ij Intensity A of light substituted for original position 1ij A first face screening picture frame 321 may be obtained. The second face screening picture frames 322, … can be obtained by the method from the second frame picture 132, and finally the face screening picture frame 32 obtained after the processing is obtained, thereby forming the processed video image stream 31. The processing procedure needs to use multiple frames of pictures, so that the processed video image stream 31 has a certain time lag compared with the original video image stream 13, and the lag amount is determined by how many processing cycles are adopted for processing. The obtained video image stream of only real human skin enters the face recognition 4 for processing, the face size and the contour are judged 41, when the face size and the contour are judged to not meet the requirements, the system outputs that the recognition fails 44, when the face size and the contour are judged to meet the requirements, the system grabs the image from the video image stream 13, and the video image stream enters the conventional face recognition algorithm 42 for further processing. When the face recognition algorithm 42 determines that the signals match, the output recognition passes 43, and when the signals do not match, the output recognition fails 44.
The chopping tri-spectral skin screening imaging technology is applied to the field of face recognition, and the problem of cheating the recognition system by using a face mask, a high-definition picture and the like can be avoided with high accuracy.
Example 2
As shown in fig. 3, a face recognition system diagram based on living human face skin detection.
In applications for imaging statistics of the number of persons in a vehicle traveling on a road, an in-car person counting system based on living face skin detection is used. The system comprises four steps of chopping tri-spectrum acquisition 1, chopping spectrum signal screening processing 2, chopping tri-spectrum output 3 and in-vehicle people counting 4The parts. The chopper tri-spectral acquisition 1 comprises a chopper 11 and an industrial camera 12. The chopper 11 is provided with a first 450nm filter 111, a second 500nm filter 112, a third 550nm filter 113, a fourth 450nm filter 114, a fifth 500nm filter 115 and a sixth 550nm filter 116, the chopper 11 rotates at a frequency F, and the industrial camera 12 shoots the outside through the filters. The frame rate of the industrial camera 12 is f, so that within 1 second an f-frame video image stream 13 can be acquired, which is represented as a first frame picture 131, a second frame picture 132, …, an f-frame picture 133. Arbitrary point P on video image space ij 14 positions: at the corresponding arbitrary point P in the first frame picture 131 1ij 141 with a light intensity A 1ij Corresponding arbitrary point P in the second frame picture 132 2ij 142 has a light intensity A 2ij …, at an arbitrary point P corresponding to the f-th frame picture 133 fij 143 with a light intensity A 3ij . The video image stream 13 enters the chopped tri-spectral signal screening process 2 for processing.
Any point P in video image ij The corresponding waveform diagram 21 shows: the light intensity of the pixel position of the living human skin changes with time for a period T=1/(2F), wherein the light intensity corresponding to the 450nm filter is A 450nm The light intensity corresponding to the 500nm filter is A 500nm The light intensity corresponding to the 550nm filter is A 550nm The skin spectrum screening process 22 is used to screen images at any point: satisfy A 450nm :A 500nm :A 550nm Approximately 1:0.4: in the case of 0.9, A ’ ij =A 450nm The method comprises the steps of carrying out a first treatment on the surface of the Not satisfy A 450nm :A 500nm :A 550nm In the case of ≡ 1:0.4:0.9, A ’ ij =0. After screening treatment, the chopped tri-spectral output 3 part is processed.
A obtained by screening ’ 1ij With an arbitrary point P on the video image space ij 14 in the first frame picture 131 in the image corresponds to P 1ij Correspondingly, the light intensity A is used ’ 1ij Intensity A of light substituted for original position 1ij A first face screening picture frame 321 may be obtained. By the methodThe second face-screening picture frames 322, … are similarly obtained from the second frame picture 132, and the face-screening picture frame 32 obtained after the processing is finally obtained, thereby forming the processed video image stream 31. The processing procedure needs to use multiple frames of pictures, so that the processed video image stream 31 has a certain time lag compared with the original video image stream 13, and the lag amount is determined by how many processing cycles are adopted for processing. The obtained video image stream of only real human skin enters the in-vehicle people count 4 for processing, firstly, the area count 41 is reserved in the face screening image, then, the circulation detection is carried out on each area, in the circulation detection, the face area and outline characteristic judgment 42 and the in-vehicle position judgment 43 are used for screening, the unsatisfied area is regarded as an invalid area, the invalid area is removed from the count, after the circulation is finished, the accurate value of the number of people in the vehicle is obtained, and the accurate value is output as the number count output 44.
The foregoing is merely a preferred embodiment of the invention, and it is to be understood that the invention is not limited to the form disclosed herein but is not to be construed as excluding other embodiments, but is capable of numerous other combinations, modifications and environments and is capable of modifications within the scope of the inventive concept, either as taught or as a matter of routine skill or knowledge in the relevant art. And that modifications and variations which do not depart from the spirit and scope of the invention are intended to be within the scope of the appended claims.