电子学报
電子學報
전자학보
ACTA ELECTRONICA SINICA
2014年
4期
658-665
,共8页
周开军%桂卫华%阳春华%谢永芳
週開軍%桂衛華%暘春華%謝永芳
주개군%계위화%양춘화%사영방
矿物浮选%泡沫图像%边缘检测%模糊逻辑%局部三值模式
礦物浮選%泡沫圖像%邊緣檢測%模糊邏輯%跼部三值模式
광물부선%포말도상%변연검측%모호라집%국부삼치모식
mineral flotation%froth image%edge detection%fuzzy logic%local ternary pattern
针对一类边缘特征不明显的矿物浮选泡沫图像,提出了一种基于模糊三值模式的泡沫图像边缘检测方法。在‘0/1’二值模式基础上,增加不确定逻辑状态,构成模糊局部三值模式,以描述邻域像素灰度均值的不确定关系,同时,对邻域双向灰度差值之和进行模糊化,以描述边缘与非边缘方向的关系,联立邻域灰度关系与双向灰度差值隶属度,构造气泡边缘隶属度矩阵,依据联合隶属度的解模糊结果判决是否为边界候选像素,再根据边界候选像素集合的特征剔除非边界像素,以此得到泡沫边缘。实验结果表明,该方法能够有效地检测出气泡边缘,同时,在强噪声环境下,具有良好的鲁棒性。
針對一類邊緣特徵不明顯的礦物浮選泡沫圖像,提齣瞭一種基于模糊三值模式的泡沫圖像邊緣檢測方法。在‘0/1’二值模式基礎上,增加不確定邏輯狀態,構成模糊跼部三值模式,以描述鄰域像素灰度均值的不確定關繫,同時,對鄰域雙嚮灰度差值之和進行模糊化,以描述邊緣與非邊緣方嚮的關繫,聯立鄰域灰度關繫與雙嚮灰度差值隸屬度,構造氣泡邊緣隸屬度矩陣,依據聯閤隸屬度的解模糊結果判決是否為邊界候選像素,再根據邊界候選像素集閤的特徵剔除非邊界像素,以此得到泡沫邊緣。實驗結果錶明,該方法能夠有效地檢測齣氣泡邊緣,同時,在彊譟聲環境下,具有良好的魯棒性。
침대일류변연특정불명현적광물부선포말도상,제출료일충기우모호삼치모식적포말도상변연검측방법。재‘0/1’이치모식기출상,증가불학정라집상태,구성모호국부삼치모식,이묘술린역상소회도균치적불학정관계,동시,대린역쌍향회도차치지화진행모호화,이묘술변연여비변연방향적관계,련립린역회도관계여쌍향회도차치대속도,구조기포변연대속도구진,의거연합대속도적해모호결과판결시부위변계후선상소,재근거변계후선상소집합적특정척제비변계상소,이차득도포말변연。실험결과표명,해방법능구유효지검측출기포변연,동시,재강조성배경하,구유량호적로봉성。
Considering a class of mineral flotation froth image with uncertain edge characteristics ,a bubble image edge detec-tion method based on fuzzy ternary pattern is proposed .On the basis of ‘0/1’binary logic ,a kind of uncertain logic state is added . According to the bubble formation mechanism and its valley edge characteristics ,the fuzzy local ternary pattern algorithm is intro-duced ,which is used to describe uncertain relationship between center and neighborhood pixel average gray level .Meanwhile ,the bidirectional grayscale difference is fuzzed to describe the relationship between edge and non-edge direction .On that basis ,the neighborhood grayscale and bidirectional grayscale difference membership are combined ,and the bubble edge membership degree matrix is constructed .Thus ,the boundary candidate pixels are determined based on the defuzzification of joint membership .Accord-ing to border candidate pixels characterized ,the non-boundary pixels are removed ,which yield the bubble edges .The experimental results show that the proposed method can effectively detect the bubble edges ;meanwhile ,the algorithm has a good robustness in high level noise environment .