Reaserch On The Indicator Supervisory System Of Ship Engine Room

The detection and alarm system of engine room is one of the most important automatic equipment on ship. Traditionally, the system was managed by computer centralized method, which is contrustrated by the following processes. First, the sensor and transmitter signals on all measure points are sent to the computer interface circuit by shielded cable. Then the signals was displayed and printed after A/D conversion and data processing. The system could also alarm when necessary. But the defects of this method is that it needs a good many sensors, transmitters and cable, and requirs high quality of computer interface, the construction and maintenance workload is also great. The article adopts computer vision and pattern recognition techniques, the monitor system's head image information which was sampled by digital camera is conveyed to computer, and then the works of feature extraction, image analysis and image processing will be done, it finally achieves the display of monitoring data and the function of alarm. This method may obviously reduce the wiring cost of the engine room and enhance the reliability of monitoring system. An overall plan of hardware and software about ship engine room monitoring system is introduced in this paper, especially the following two aspects: image gathering and recognition. It mainly concerns about the following content:1. This article uses Visual C++6.0 as the development environment for the video software. By contrasting the difference of Direcshow and VFW, the design adopts the VFW to develop the procedure of video capture, which is flexible, reliable and speedy. It also makes the capture procedure much more independent of the video equipment. Because of the supporting of data stream processing, the VFM makes the gathering of real-time data and the demonstration of monitoring picture come true.2. When interception and processing the video image, the multi-thread programming technology ensures the smooth demonstration of the video monitoring picture while processing the frame. In order to fulfill the requirement of time and efficiency, the design introduces direct memory store into frame capture of the video stream, it builds a good foundation for the following image processing.3. The system carries on gradation processing to the color image in pretreatment. Considering the environment of engine room and the character of image of apparatus indicator, by comparing the different algorithms, the design employs value filter to the gradation image to eliminate the noise and obtain the complete image and edge information, it also use Ostu to bring the binary image. For the sake of identifying the angle of the indicator faster and more exactly, the image has been thinned. All these pretreatment takes good prepare for the extraction and identification of image.4. The article has compared several methods to detect the indicator to identify the angle. Considering the stability, rapidity and accuracy of the supervisory system, the paper chooses the Hough transformation, it contains not only the principle and improvement of the HT, but also the extraction of indicator and the identification of the angle which are completed by MHT.