Study On Mobile Monitoring And Recongnization Methord Of Finite-state Devices

In many industrial fields, the working state of chief devices need to be monitored for the reason of security.Usually, the quantization picture of the state of device is sent to the center of monitoring, then operator or computer analyse and judge the state of device. Howerver, some state of device is not quantized and the security of sowme device is not predicted, so the function of monitoring is not substituted.With the development of economy and socity, the system of monitoring has used to many fieldes in socity.Nowadays, many monitoring systems are fixed. However, many places need more monitoring system, such as power substation. If we build and maintain these systems, it will cost big huge fund and manual labour. If we use mobile monitoring system that install the video monitoring onto a handcart called mobile monitoring device (MMD). MMD patrol every target device according to the ordered route and recognize the real-time state of target device. By recognize the sign attached to target device, MMD can find the target device and send picture to center of controlling. Compare to fixing monitoring system, MMD has some advantages: First, it has less workload. Second, it cost small fund and manual labour. But there are some disadvantages in this system. First, there is big Channel transmission pressure. Second, it still needs many human monitoring work.Therefore, this paper proposed an approach of moving monitoring and real-time recognition. MMD analyse and recongnize picture of device that are monitored by MMD. Some state of device that is not quantized must be patrolled for the normal working state by operator in person, so that can ease human work and chanell transmission pressure.Because hundreds of device that have different size,shape and color need to be monitored by MMD. This paper recongnizes the status of device by judging the similarity between real-time image and standard image. This method requires capture the same size and visual angle between real-time image and standard image.To solve this problem this paper add a special sign on device, this sign also is used tofix positon.This paper uses the approach to recognize the state of target device. Themain results of this study can be summarized as follows:(1) An approach for recognizing symbols based on the characteristics ofsymbolic heart-shaped was proposed, after scanning the image once and theconnected-domain queue twice, the symbol-domains waiting for being identified canbe extracted, combining with the number and heart-shaped distribution feature ofconnected-domains in the symbol-domains, symbols can be recognized.(2) For collecting clear image of the state of device, it needs to focus motor ofcamera. So after studying on several focusing algorithm, this paper propose a fastfocusing algorithm based on dicholomy.(3) Study on algorithm of controlling PTS and controlling zooming. ControllingPTS is a necessary factor of video monitoring. But the previous controlling PTSsystem can not achieve to the criterion. So we must improve the performance of theold PTS. Then, Using the information of sign that has been recognized, this paperpropose a algorithm of controlling PTS based on sign and a algorithm of controllingzooming based on sign.(4) Study on recognition algorithm of finite state of device. By study deeply onmany kinds of methods of recognition, this paper implements several recognizer offinite state of equipment. At last it proposes a method of novel recognizer based onvote mechanism.