激光技术
激光技術
격광기술
LASER TECHNOLOGY
2013年
5期
690-695
,共6页
高颖%王阿敏%王凤华%郭淑霞
高穎%王阿敏%王鳳華%郭淑霞
고영%왕아민%왕봉화%곽숙하
图像处理%图像融合%边缘检测%小波变换%图像评价
圖像處理%圖像融閤%邊緣檢測%小波變換%圖像評價
도상처리%도상융합%변연검측%소파변환%도상평개
image processing%image fusion%edge detection%wavelet transform%image evaluation
为了改善传统图像融合方法在精确制导武器系统中对目标检测模糊、识别率低与实时性差等缺陷,采用了一种将小波变换与Canny算子相结合的图像融合的新方法。该方法的具体改进在于首先对源图像在垂直和水平方向上进行了适合图像重构的3层小波分解,并依据各分解层不同频率分量的自身特性,采用独特的融合规则,即对低频分量采用加权平均融合算法,对高频分量采用Canny 算子与局部区域均方差准则结合法改变图像的小波系数,最后对融合后的小波系数进行逆变换,得到重构的目标图像。结果表明,利用该方法不仅降低了融合图像的边缘模糊性,突出了目标色彩,达到良好的视觉效果,而且计算效率高、实时性好,特别有助于伪装目标的检测与识别,具有较好的应用价值。
為瞭改善傳統圖像融閤方法在精確製導武器繫統中對目標檢測模糊、識彆率低與實時性差等缺陷,採用瞭一種將小波變換與Canny算子相結閤的圖像融閤的新方法。該方法的具體改進在于首先對源圖像在垂直和水平方嚮上進行瞭適閤圖像重構的3層小波分解,併依據各分解層不同頻率分量的自身特性,採用獨特的融閤規則,即對低頻分量採用加權平均融閤算法,對高頻分量採用Canny 算子與跼部區域均方差準則結閤法改變圖像的小波繫數,最後對融閤後的小波繫數進行逆變換,得到重構的目標圖像。結果錶明,利用該方法不僅降低瞭融閤圖像的邊緣模糊性,突齣瞭目標色綵,達到良好的視覺效果,而且計算效率高、實時性好,特彆有助于偽裝目標的檢測與識彆,具有較好的應用價值。
위료개선전통도상융합방법재정학제도무기계통중대목표검측모호、식별솔저여실시성차등결함,채용료일충장소파변환여Canny산자상결합적도상융합적신방법。해방법적구체개진재우수선대원도상재수직화수평방향상진행료괄합도상중구적3층소파분해,병의거각분해층불동빈솔분량적자신특성,채용독특적융합규칙,즉대저빈분량채용가권평균융합산법,대고빈분량채용Canny 산자여국부구역균방차준칙결합법개변도상적소파계수,최후대융합후적소파계수진행역변환,득도중구적목표도상。결과표명,이용해방법불부강저료융합도상적변연모호성,돌출료목표색채,체도량호적시각효과,이차계산효솔고、실시성호,특별유조우위장목표적검측여식별,구유교호적응용개치。
In order to overcome the defects of fuzzy detection , low recognition rate and poor real-time of traditional fusion methods used in precision-guided weapons systems , a new image fusion algorithm was proposed combining wavelet transform with Canny operator .Firstly, the source image was decomposed into 3 layers in vertical and horizontal directions , which are suitable for image reconstruction; then due to its own characteristics of the different frequency components , an unique fusion rule was used to change wavelet coefficients of images , that is, for the low frequency components , the weighted average fusion algorithm was adopted , and for the high-frequency components , wavelet coefficients were changed using Canny operator and the local area variance criteria method .Finally, images were reconstructed using the inverse wavelet transform for different components .Results show the improved method not only reduces the fuzziness of edge , highlights target color , gets better visual effects , but also makes computational efficiency high , real-time good, particularly can detect and recognize pretend targets .It has better theoretical research and application value .