烟草科技
煙草科技
연초과기
TOBACCO SCIENCE & TECHNOLOGY
2015年
1期
96-100
,共5页
烟草%异物剔除%低照度%图像增强%同态滤波%实时性
煙草%異物剔除%低照度%圖像增彊%同態濾波%實時性
연초%이물척제%저조도%도상증강%동태려파%실시성
Tobacco%Foreign matter rejection%Low illumination%Image intensification%Homomorphic filtering%Real-time
为解决烟草异物剔除系统随照度衰减后系统剔除性能下降问题，采用优化的高斯同态滤波算法在系统中增设了图像增强功能。先将低照度烟叶图像从RGB（Red-Green-Blue）空间快速转换到HSV（Hue-Saturation-Value）空间，实现色彩与亮度的分离，采用实时性较高的空域同态滤波方法对亮度分量V进行增强，引入自适应系数对亮度进行拉伸，最后将HSV转换到RGB模式。结果表明，该方法能有效校正低照度图像的颜色、对比度和亮度，对光照不均有很好的均衡作用，具有较好的自适应性；能够较好地保持系统剔除性能，增强系统的可维护性、易操作性；与其他图像增强算法相比，本方法运算速度更快，能较好地满足实时增强彩色图像的需求。
為解決煙草異物剔除繫統隨照度衰減後繫統剔除性能下降問題，採用優化的高斯同態濾波算法在繫統中增設瞭圖像增彊功能。先將低照度煙葉圖像從RGB（Red-Green-Blue）空間快速轉換到HSV（Hue-Saturation-Value）空間，實現色綵與亮度的分離，採用實時性較高的空域同態濾波方法對亮度分量V進行增彊，引入自適應繫數對亮度進行拉伸，最後將HSV轉換到RGB模式。結果錶明，該方法能有效校正低照度圖像的顏色、對比度和亮度，對光照不均有很好的均衡作用，具有較好的自適應性；能夠較好地保持繫統剔除性能，增彊繫統的可維護性、易操作性；與其他圖像增彊算法相比，本方法運算速度更快，能較好地滿足實時增彊綵色圖像的需求。
위해결연초이물척제계통수조도쇠감후계통척제성능하강문제，채용우화적고사동태려파산법재계통중증설료도상증강공능。선장저조도연협도상종RGB（Red-Green-Blue）공간쾌속전환도HSV（Hue-Saturation-Value）공간，실현색채여량도적분리，채용실시성교고적공역동태려파방법대량도분량V진행증강，인입자괄응계수대량도진행랍신，최후장HSV전환도RGB모식。결과표명，해방법능유효교정저조도도상적안색、대비도화량도，대광조불균유흔호적균형작용，구유교호적자괄응성；능구교호지보지계통척제성능，증강계통적가유호성、역조작성；여기타도상증강산법상비，본방법운산속도경쾌，능교호지만족실시증강채색도상적수구。
The foreign matter rejecting performance of a tobacco sorting system decreases with the attenuation of illumination, therefore the function of image intensification was incorporated on the basis of optimized Gauss homomorphic filtering algorithm. The image of tobacco leaf under low illumination was converted swiftly from its RGB (Red-Green-Blue) space into an HSV (Hue-Saturation-Value) space in order to separate color from luminance, and the luminance component V was intensified by a real-time spatial homomorphic filter, then stretched by an introduced adaptive coefficient, finally HSV was converted into RGB. The results showed that this method could effectively correct the color, contrast and luminance of an image under low illumination, well balance uneven illumination. The modified system maintains its good rejecting performance, features better maintainability and operational convenience. Comparing with other image intensification algorithms, this method features better self-adaptability and faster processing speed and is more applicable to real-time intensification of color images.
