【摘要】：為滿足手指交互系統(tǒng)的需要,并能夠達到對傾斜指勢進行準(zhǔn)確識別的要求,本文介紹了一種快速、準(zhǔn)確對指尖檢測定位并實時識別傾斜指勢的方法。該方法利用YCb Cr顏色空間分割算法對膚色聚類進行粗分割,然后運用"周積比"概念對預(yù)處理區(qū)域進行細(xì)化分割,剔除除手部以外的膚色干擾區(qū)域并利用最小二乘法二項式擬合算法獲取手指輪廓。采用改進的凸包絡(luò)優(yōu)化算法完成指尖的檢測及傾斜修正。最后,進行局部掃描獲得最終的指勢識別。實驗表明本文介紹的方法能實現(xiàn)簡單傾斜指勢0?9的識別且識別率高達95.7%,穩(wěn)定性較好。
[Abstract]:In order to meet the needs of finger interaction system and to recognize the tilted finger potential accurately, this paper introduces a fast and accurate method to detect and locate the finger tip and identify the tilting finger potential in real time. In this method, YCb Cr color space segmentation algorithm is used for rough segmentation of skin color clustering, and then the concept of "circumference ratio" is used to refine and segment the preprocessed region. The skin color interference area except the hand is eliminated and the finger contour is obtained by using the least square binomial fitting algorithm. The improved convex envelope optimization algorithm is used to detect and correct the fingertips. Finally, the final finger potential recognition is obtained by local scanning. The experimental results show that the method presented in this paper can realize the recognition of simple tilting finger potential 0 / 9 and the recognition rate is as high as 95.7 and the stability is good.
【作者單位】： 上海大學(xué)新型顯示技術(shù)及應(yīng)用集成教育部重點實驗室;上海大學(xué)微電子研究與開發(fā)中心;
【基金】：國家自然科學(xué)基金(61376028)資助項目
【分類號】：TP391.41
，

本文編號：2293150