Abstract: In the application of visual detection of longitudinal tearing of coal mine conveyor belts, the visual system is prone to problems such as dust blockage and contamination, which affect the clarity and accuracy of image acquisition. At present, the continuous cleaning method using a glass housing with a wiper can maintain image availability, but the accuracy of data analysis is affected due to the low quality of collected images. To address the above problems, a design scheme of a dust removal system based on an intermittent rotary glass component and brush was proposed. A 36BL57-1242 Direct Current (DC) brushless motor was used to drive a gear system to rotate the glass component and the brush separately to remove coal dust from the short glass cover of the laser emitter and the long glass cover of the industrial camera. Based on the STM32F030F4P6 Microcontroller Unit (MCU) chip, precise motion control of the DC brushless motor was realized, including start-stop, steering, Pulse Width Modulation (PWM) speed regulation, and overcurrent protection. When overcurrent or stalling of the motor was detected, a shutdown alarm was immediately triggered. During the dust removal process, considering that continuous operation would lead to continuous contact of the bristles with dust and weaken the rebound cleaning ability of the brush body, an intermittent operation mode was adopted. A magnet was embedded into the intermediate gear and combined with a Hall switch to form periodic induction, and a dual-voltage comparator was used to construct a position-feedback closed loop, thereby realizing the periodic action of the motor driving the glass component to perform a full rotation and then stop for 3 seconds, which facilitated rebound dust removal. Laboratory and underground test results showed that during 2 000 hours of intermittent rotary dust removal operation, the motor operated stably. The images obtained with laser irradiation on the conveyor belt had high contrast, with no occurrences of dim laser lines, blurred images, or unrecognizable conditions caused by contamination. The intermittent dust removal design effectively cleaned the coal dust on the glass cover, ensured accurate image recognition, and thus obtained reliable tearing detection results.