The Research On Key Technology Of Ship Fire-smoke Detection In Maritime Video

Fire hazard is a common one among the inland waterway traffic safety accidents.Ship fire hazard does not only cause serious destruction on the safety of state property,but also do damage to the local maritime environment simultaneously. Therefore, inthe field of intelligent monitoring, the fire smoke detection surrounding around theship engine room and the hull is a very important link. Has important theory researchsignification and utility value.During the early fire period, it usually produces a small amount of smoke ofdifferent degree of concentration of color, for detect the ship fire as early as possible;this paper mainly studies the method of ship fire smoke detection and some relatedalgorithms. According to the navigation environment of inland waterway transportsystem, we propose a solution which contains the motion region extraction andmotion detection, the smoke feature recognition and intelligent alarm three steps.In terms of moving region detection, this paper mainly focuses on the analysis ofonline gray clustering method based on background reconstruction. In the backgroundmodel learning stage, the fluctuations on the water and background disturbances willcause the wrong background reconstruction model, then it will take so much time torebuild the background model, so, we put forward a kind of adaptive threshold andcumulative weighted and multi-background reconstruction method. First, all pixelsare divided into static and non-static points; Then, Only build multi-background onthe non-static points. This method improves the accuracy of background model,narrowed the background modeling time.In terms of smoke detection, this paper analyzes the traditional cumulativedirection based on the movement and smoke recognition method based on mix theinformation of color and energy. In view of the smoke point mobile cases, weproposed a smoke detect method base on the fusion of dynamic motion characteristics,color information and image gray matching. First, remove interference objects usingdynamic motion characteristics; then filtering the image element through the colorinformation and image matching features; Finally, through the comprehensivestatistics in time and space, and obtain the smoke recognition results. This algorithm improves the accuracy and the effectiveness of smoke recognition results under thecomplex environment.Finally, based on VS2010platform with OpenCV library, this paper carries outthe ship smoke detection algorithms in maritime image sequence. We havelarge number of experiments in different videos, the experimental results show that:this method cans effectively detect the smoke signals in different environment andachieve the purpose of early fire smoke monitoring and alarming.