Abstract: The color,position and width of mineral belts on the table concentrator often change due to the variations of feed capacity and pulp density,then operators need to adjust the partition plate position in time to reach the qualified concentrate grade which is labor-intensive.The table concentrator mineral belt images were sampled by machine vision and the mineral belt images recognition algorithm was studied in order to realize automatic inspection and operation of table concentrators.On this basis,an intelligent control system was developed.Industrial experiments show the automatic inspection robot can realize continuous sampling of mineral belt images from a group of table concentrators.The image recognition algorithm can exactly obtain the width,boundary and color features of the mineral belts and the control system can automatically adjust the plate to the target position,thus achieving the goal of replacing manual labor and improving the ore-dressing index.