结合多尺度边缘增强及自适应谷底检测的浮选气泡图像分割

针对浮选气泡图像噪声大、边界弱、传统谷底检测算法对不同类型气泡分割不具普遍性等问题,提出了一种结合Contourlet多尺度边缘增强及自适应谷底边界检测的气泡分割方法。该方法通过对气泡图像进行Contourlet分解,得到多尺度多方向高频子带;通过对各方向子带的高频系数进行非线性增益处理,实现边缘增强和噪声抑制。对和声搜索算法的"调音"策略和参数设定方法进行了改进,对不同类型气泡图像自适应地获取谷底边界检测算法的最优参数,提取谷底并进行形态学的边缘完善处理。最后进行了分割实验,并与其它方法做了比较。结果表明,采用该方法对不同类型气泡进行分割时,平均检测效率(DER)和准确率(ACR)分别为91.2%和90.6%,较传统分割方法有较大提高。该方法无需手工调节参数,自适应能力强,精度高。

Flotation froth image segmentation based on multiscale edge enhancement and adaptive valley detection

To overcome the weak edges and large noise of flotation froth image, and to solve the weakness of traditional valley detection algorithm on different kinds of bubble segmentation sizes, a froth image segmentation method was proposed based on Contourlet transform multi-scale edge enhancement and adaptive valley detection. Firstly, the froth image was decomposed by using the Contourlet transfom to obtain multi-scale and multi-direction sub-band coefficients. Then, thresholds of the nonlinear enhancement function were determined according to the coefficients of each scale to enhance edges and suppress the noise. Furthermore, the optimal position adjustment strategy and parameter setting of HS were improved to find the optimal parameters of valley detection algorithm and to detect the different kinds edges of bubble image size. Finally, segmentation experiment was performed and obtained result was further improved by morphological processing. Experiments show that the proposed method effectively detects the edges of different type of bubbles adaptively, and the average detection efficiency (DER) is 91.2% and the average accuracy (ACR) is 90.6%, which is much better than that of traditional methods. This method has high precision, good adaptive ability, and does not need to adjust parameters manually.