표적 영상 중심추적 기반 LUT를 활용한 열 화상 카메라의 줌 렌즈 구동 흔들림 보상기법

김병학; 김민영

doi:10.5302/J.ICROS.2017.16.0189

주제분류

...

저널정보

저자정보

김병학 (경북대학교)
김민영 (경북대학교)

이용수
내서재: 0

내서재에 추가
되었습니다. 내서재에서
삭제되었습니다.

초록·키워드

오류제보하기

The magnification of infrared and normal tracking cameras for advanced aircraft and autonomous vehicles need to be continuously adjusted using a lens driving function to define target objects and keep track of them. However, they still suffer from the problem of retaining stability of the focused image even when a continuous zoom lens driving situation is in operation. According to the mechanical motion characteristics of a magnification lens, the motorized movement of the lens causes an incremental spiral motion on the image along the direction of movement. In addition, the optical alignment error in the zoom lens, moving stages, and imaging area seriously amplifies this problem. The spiral movement of the video image is clearly observed on a user monitor if it is in focus during zooming in or out. The spiral motion is more clearly observed under high magnification than under low magnification, so serious inconvenience is caused for users at high magnification. To solve this problem, an image-based compensation algorithm based on quantitative measurement of spiral image movement in zoom driving was investigated. First, image movement was measured and analyzed under continuous zoom lens driving conditions with a standard target image and a collimator system based on a centroid tracking algorithm. Second, a proposed motion compensation technique using a look-up table (LUT) constructed from the measured movements was applied. To prove the validation of this technique, a series of experiments were performed on a real forward looking infrared (FLIR) image sensor system with a zoom lens structure. It was confirmed that the technique can significantly reduce the spiral movement induced by zooming in or out to ±1 pixel.

#thermal camera #zoom lens #4-bar target #centroid tracking #look-up-table control method #LOS (Line of Sight) #image stabilization