CAJ | 학술논문

　　针对目前指尖检测中存在漏检、误检且易受手区域尺寸大小影响等缺陷，提出了一种加强检测性能的指尖检测框架，对算法自适应性进行了针对性的改进。在利用运动信息和肤色信息提取出手区域的基础上，用包含排漏算法的自适应指尖检测算法检测出指尖，通过提取每个指尖对应的指宽、指长、指向以及手掌宽度信息对检测到的指尖进行筛选，排除误检的指尖，分别与固定k-余弦算法和可变k-余弦算法进行比较，实验结果表明提出的自适应指尖检测算法检测率高、鲁棒性好。
　　침대목전지첨검측중존재루검、오검차역수수구역척촌대소영향등결함，제출료일충가강검측성능적지첨검측광가，대산법자괄응성진행료침대성적개진。재이용운동신식화부색신식제취출수구역적기출상，용포함배루산법적자괄응지첨검측산법검측출지첨，통과제취매개지첨대응적지관、지장、지향이급수장관도신식대검측도적지첨진행사선，배제오검적지첨，분별여고정k-여현산법화가변k-여현산법진행비교，실험결과표명제출적자괄응지첨검측산법검측솔고、로봉성호。
A robust system framework of fingertips detection is proposed to deal with missing detections, false detections, and sensitivity to the hand area size. After detection of the hand area based on motion information and skin color, fingertips are detected by an adaptive fingertips detection algorithm which can avoid missing detections. Then the parameters such as finger width, finger length, finger orientation, and palm width are used to filter the false detections.The experimental results show that the proposed adaptive fingertips detection algorithm has high detection rate and good robustness compared with the existing fixed and variable k-cosine algorithms.