The Key Problem Of Coal Bunker Detection Based On Computer Vision

In the coal production enterprises, we always set a certain amount of coal bunker to ease the storage and transit in the process of coal transportation. Therefore, the depth detection of coal bunker is the key measurement to avoid bunker short position caused by a short circuit and coal overflow cause the equipment accident. It is the important condition for the coal enterprise security and efficient operation. However, due to the coal varied, high dust, high humidity, low light and gas gathering of the coal bunker’s harsh environment restricts the development and application of coal bunker smart detection and coal technology severely, so that it is difficult to meet the needs of coal mine production safety. Therefore, the position measurement of coal bunker is a technical problem unsolved for a long time. In the paper, firstly,we introduced a actual statue of coal bunker and analyzed and summarized the exist coal bunker detection methods and coal bunker structure and disadvantage based on the analysis of the contact and non-contact coal level detection. The exist self-calibration algorithm and the process of calibration were studied included in the two-step, self-calibration and the plane calibration algorithm. Theoretical model for coordinate transformation carried out a detailed analysis. Through the simulation experiment proved that the proposed adaptive calibration of coal bunker detection is effectiveness and feasibility. According to the exist calibration algorithm, the results of coal bunker detection we proposed were compared and analyzed. The calibration algorithm we proposed does not require any calibration templates and is simple. The time can be saved in the later part of image processing and real-time three-dimensional reconstruction of the latter provides a large amount of data support. Due to the harsh environment such as the severe dust, large moisture, low light image,we proposed a new algorithm based on the Gaussian pyramid algorithm and Mallat algorithm by analyzing the basic characteristics of the camera imaging in coal bunker detection to guarantee the implementation of coal bunker visual inspection smoothly and effectively. Taking into account the imaging environment in the coal bunker and the particular of coal bit, the algorithm combines the Mallat algorithm and Gaussian pyramid algorithm to reconstruct and analyze the coal bit image in the multi-scale space and remove the redundant information maximize. Feature information optimized can be reconstruct to guarantee the reconstruction of coal bit image timely. The test was apparent to show the algorithm we proposed in this paper can get the actual three-dimensional position information of the key points in the coal bunker, thus, a controllable intelligent forecasting of coal bit in coal bunker can be achieved and it is the foundation for coal enterprise security, stability and effective conduct.