论文部分内容阅读
矿物浮选的精矿品位与泡沫的纹理复杂度密切相关.为了避免纹理谱方法中像素值的刚性比较,提出了一种新的基于旋转分类模糊纹理谱的纹理描述方法用于金属矿物分离.经过指数函数拟合频率分布曲线,确定模糊阈值大小,统计模糊纹理单元,给出了灰度差值的模糊纹理谱,并利用旋转分类将6561种纹理单元简化至834种.基于此,引入2个描述因子,纹理平滑度与粗糙度.实际应用结果表明,纹理平滑度可描述图像纹理粗糙程度,同时反映实时矿物品位.
In order to avoid the rigid comparison of pixel values in texture spectrum method, a new texture description method based on rotational classification fuzzy texture spectrum is proposed for metal mineral separation. After fitting the frequency distribution curve by exponential function, the fuzzy threshold size was determined, the fuzzy texture unit was calculated, and the fuzzy texture spectrum of the gray difference value was given, and the 6561 kinds of texture units were simplified to 834 by rotation classification. Describing factor, texture smoothness and roughness, the practical application results show that the texture smoothness can describe the roughness of the image texture and reflect the real-time mineral grade.