Resumen
[0001] A system (100) for vein location and vascular biometric recognition using mobile devices (110), such as a smartphone (111), a laptop (112), a tablet (113), an embedded system (114) and/or any portable smart device (115), the system (100) comprising a contactless device (101), connected with the mobile device (110) via a USB interface (130), with one or two near-infrared cameras (103) and two diffuse pulsed near-infrared LED spotlights (104). The near-infrared cameras (103) and the near-infrared LED lights (104) communicate with a processing unit (102), which executes Al-based algorithms using a convolutional generative adversarial network (500, 600, 700) for infrared image processing and enhancement, vein visualization, real-time guidance of health professionals, and automatic vascular biometric recognition. The network (500, 600, 700) obtains 2D or 3D images of vein patterns, which can be displayed on the mobile device (110). The 3D reconstructions of vein depth can be obtained from i) input images using a stereo pair infrared cameras, ii) input information of Time-of-Flight or iii) both input at the same time.
Reivindicaciones
1. A system (100) for locating veins in a human or animal subject using a mobile device (110), characterized by comprising: - a contactless device (101) pluggable into the mobile device (110) via a universal serial bus interface (130), the contactless device (101) integrating one or two near-infrared cameras (103) for capturing images of vascular lines and surrounding tissue of a subject's body part in real-time; and - a processing unit (102) communicated with each near-infrared camera (103), the processing unit (102) being configured to: - receive (310, 410) the captured images from each near-infrared camera (103), - preprocess and enhance (320, 420) the received images by a convolutional generative adversarial network, CGAN (500, 600, 700), which applies artificial intelligence for processing two-dimensional images (321, 421), if only one near-infrared camera (103) is used, and three-dimensional images (322, 422), if two near-infrared cameras (103) are used, the CGAN (500, 600, 700) obtaining two-dimensional or three-dimensional images of a vein pattern of the subject's body part; - send in real-time the obtained images of the vein pattern to be displayed (340, 440) by using a graphic user interface provided by the mobile device (110); - provide a user of the system (100) with an indication of a result from an advisory unit (350, 450) integrated into the processing unit (102), the advisory unit (350, 450) obtaining the result, which is related to the location of veins in the vein pattern, by using deep learning based on a vision transformer and a convolutional recurrent neural network. 2. The system (100), according to any preceding claim, wherein the contactless device (101) integrates two near-infrared cameras (103) and further comprises two diffuse pulsed near-infrared light-emitting diodes (104) communicated with the near-infrared cameras (103) and the processing unit (102). 3. The system (100), according to claim 2, wherein the CGAN (600, 700) of the processing unit (102) is configured to obtain three-dimensional images of the vein pattern by using each pulsed near-infrared light-emitting diode (104) associated with each near-infrared camera (103) as a time-of-flight sensor (324, 424) and combining the captured images from the two real-time dual infrared cameras to estimate (620, 630, 720, 730) a depth of each vein line in the vein pattern. 4. The system (100), according to claim 3, wherein the processing unit (102) is further configured to filter the obtained images of the vein pattern by flow blood computed using the vein depth/section estimations (620, 630, 720, 730) from the CGAN (600, 700). 5. The system (100), according to any preceding claim, wherein the CGAN (500) of the processing unit (102) is configured to enhance (320, 420) the received images by increasing the contrast between vascular lines and surrounding tissue in the images to be displayed. 6. The system (100), according to claim 5, wherein the CGAN (500) of the processing unit (102) is configured to output non-hair and high-contrast enhanced images (515). 7. The system (100), according to any preceding claim, wherein the processing unit (102) is configured to send the images to be displayed (340, 440) in real-time by using augmented reality or mixed reality displayable by the mobile device (110) or on a screen of the mobile device (110). 8. The system (100), according to any preceding claim, wherein the processing unit (102) is configured to send the images to be displayed (340, 440) by video streaming. 9. The system (100), according to any preceding claim, wherein each near-infrared camera (103) has autofocus and a resolution of 2160 pixels, 1080 pixels, 720 pixels, or 480 pixels at 30 frames per second, 60 frames per second, or 120 FP frames per second. 10. The system (100), according to any preceding claim, wherein the CGAN of the processing unit (102) is configured to use additional infrared images of vein patterns obtained from external Big Data systems. 11. The system (100), according to any preceding claim, wherein the processing unit (102) is further configured to record and store (330) the obtained images to train the deep learning of the advisory unit (350, 450). 12. The system (100), according to claim 9, wherein the advisory unit (350) is configured to indicate a puncture area for the subject based on the deep learning training from the obtained images of the vein pattern for the subject and a group of subjects, or the advisory unit (450) is configured to indicate an identification of the subjects based on the vascular biometric reconstruction performed by the deep learning training from the obtained images of the vein pattern. 13. The system (100), according to any preceding claim, wherein the mobile device (110) is a smartphone (111), a laptop (112), a tablet (113), a desktop computer, a wearable gadget, or an embedded system (114).
