基于PSO-Faster R-CNN改进算法的矿石识别分类研究

Research on Ore Identification and Separation Based on Improved PSO-Faster R-CNN Algorithm

为提升矿物的分选效率,基于机器视觉对矿石识别分选算法进行研究。采用PSO-Faster R-CNN改进算法对矿石二维图像进行识别分类,PSO算法改进了Faster R-CNN算法的收敛速度,使用了多尺度特征融合的方法,实现了对特定矿石图像样本的识别。此外,还选择了HOG+SVM算法和KNN算法进行对比。结果表明,PSO-Faster R-CNN改进算法识别精确度高达98%,精度和稳定度优于其他两种算法,实际工程识别速度快,未来能满足矿石分选识别的应用需要。

In order to improve the separation efficiency of minerals,the ore identification and separation algorithm was studied based on machine vision.The improved PSO-Faster R-CNN algorithm was used to identify and classify the two-dimensional ore images.The convergence speed of the Faster R-CNN algorithm was improved by PSO algorithm,and the multi-scale feature fusion method was used to realize the identification of specific ore image samples.In addition,HOG+SVM algorithm and KNN algorithm were selected for comparison.The results show that,the identification accuracy of the improved PSO-Faster R-CNN algorithm is as high as 98%,and the accuracy and stability are better than those of the the other two algorithms.The identification speed in actual engineering is fast,which can meet the application needs of ore separation identification in the future.