Abstract:
본 발명의 한 실시예에 따른 진단 위치 추적 장치는 환자의 얼굴 영상을 촬영하거나, 환부를 관찰하는 진단기기에 부착된 마커(marker)를 촬영하는 촬영부, 조작자로부터 환자 정보 또는 설정 정보를 입력 받는 입력부, 열굴 촬영 영상 또는 마커 촬영 영상을 처리하며, 상기 얼굴 촬영 영상의 특징점과 상기 마커의 위치 관계를 추출하는 제어부, 상기 얼굴 촬영 영상의 특징점, 상기 마커의 위치 관계 및 상기 환자 정보를 저장하는 저장부, 그리고 상기 환자 정보를 입력 받기 위한 프레임, 상기 설정 정보를 입력 받기 위한 프레임, 상기 얼굴 촬영 영상을 출력하기 위한 프레임 및 상기 마커 촬영 영상을 출력하기 위한 프레임 중 적어도 하나를 표시하기 위한 그래픽 유저 인터페이스(Graphic User Interface, GUI)를 제공하는 출력부를 포함한다.
Abstract:
Provided is a nose region detection method capable of accurately detecting a nose region in visible rays environment where illumination changes in a various way as well as an even infrared rays environment and accurately detecting a nose region even when the two nostrils are not found in a nose region depending on a pose of a user. The nose region detection method extracts a face area and facial characteristics in the face area from the acquired face image and extracts a nose search area based on the extracted face area and facial characteristics in the face area. Then, the nose area is detected through an integral imaging method and one or more nostrils shape characteristics. And the nose area is detected through an adaptive template matching and the nose search area of a next frame image in the current frame image to track the nose area in a sequence images.
Abstract:
A device for recognizing a face comprises: a communications interface which receives a face image; a processor which identifies the face of a user showed on the face image according to a command; and a memory which stores the command. The command comprises: a step for dividing a face box area of the face image using a central vertical line, and setting one side area and the other side area; a step for comparing the one side area with the other side area and determining whether or not the face of the face box area is rotated; a step for rotating the face box area around a center point of the face box area at a predetermined angle when the face is rotated; and a step for identifying the face of the user using the face box area when the face is not rotated.
Abstract:
The present invention relates to a method and apparatus for identifying money. The method according to the present invention includes the steps of: obtaining a money image which is inputted; analyzing the money image by applying a subspace analysis method to the image data of the obtained money image; and identifying the money image based on am analysis result. [Reference numerals] (101) Image generation unit; (102) Interest region extraction unit; (110) Image contraction unit; (120) Analysis unit; (130) Memory unit; (140) Determination unit
Abstract:
PURPOSE: A medical position tracking device is provided to diagnose by taking a picture of a skin and a scalp state with an identical view angle about an affected part of a patient. CONSTITUTION: A rotating unit (200) is combined in an upper part of a support stand (100). A frame (110) is formed in a lower part of the support stand. A face fixing unit (300) is installed at the support stand, and fixes a face of a patient. A first shooting unit (310) is positioned at one side of the face fixing unit, and takes a picture of the face of the patient. A location tracking unit (210) comprises a second shooting unit (320) taking a picture of an affected part image of the patient. The location tracking unit rotates around the rotating unit.
Abstract:
PURPOSE: A focus measurement apparatus of an eye tracking system using multilayer neural network is provided to assemble the capability of collecting high frequency elements in an image based on a spatial region and the capability of reducing the influence of brightness change of an image based on a wavelet region by using the multilayer neural network. CONSTITUTION: An image acquisition unit (1105) acquires an image of an eye by using a narrow angle camera. An input image generation unit (1110) generates an input image by down sampling of the image of the eye. A brightness compensation unit (1115) compensates brightness of the input image. A normalization unit (1125) normalizes four focus values measured in a focus measurement unit (1120) nonlinearly. A focus value addition unit (1130) adds up the normalized four focus values into one final focus value through multilayer neural network. A lens control unit (1135) controls a focus lens by using the final focus value. [Reference numerals] (1105) Image acquisition unit; (1110) Input image generator; (1115) Brightness compensation unit; (1120) Focus measurement; (1125) Normalization unit; (1130) Focus value addition unit; (1135) Lens controller