The Study Of The Oil Spills Detection And Classification Based On SAR Images

Oil spill can lead to seriously environmental damage and property loss. Effectiveoil spill detection and monitoring is the basis for the rapid response. It has developeda wide variety of oils pill detection methods, in which the best methods is SAR imagedetection.This paper mainly researched the extraction and the classification of oil spill inSAR image. The radar detection of marine oil spill always uses the characteristics ofits Sensitive reaction to the surface roughness of ground targets. After the oil spill, oilfilm spread to sea surface produces a damping effect to gravity capillary waves of seasurface. It makes surface which spread by oil film smoother than the usual, then thebackscattering radar echoes became smaller. So the oil film can extract by collectlower scattered areas on image namely the dark areas.This paper uses some image segmentation methods to extract the oil filminformation, such as single threshold, texture information, level set, etc. The methodof level set as a kind of edge detection method is popular in recent years. When thismethod was used in edge detection, it does not rely on the local information of theimage. Thus in the fuzzy edge detection it has a very good detection effect.Many marine phenomena in SAR images have shown the low dark backscatterregion, because the SAR image backscatter dark area has multi‐solutions, It is usedthe information, the texture of SAR images using artificial neural network intelligentidentification ways to classify the oil slick, exclude the suspected oil slick.This study using statistical methods based on gray level co‐occurrence matrixtexture features extracted from SAR image values, selects five texture index featuresof the mean, entropy, homogeneity, contrast and angular second moment as BPneural network input to train the network; Then use the trained network to completethe classification of the each pixel of the image, finally the effect of the classificationis simply evaluated. The characteristic of intelligence distinguishing function ofmanpower neural network makes the oil slick using this method to classify, which isacquired better results.The paper also briefly introduces the oil spill extraction process and thedistinguishing of the pollution oil film and the leakage oil film. Using the SAR imageand multi‐source remote sensing image to detect the oil spill, with the help ofmarine facilities and geological faults and gravity and magnetic anomalies andgeochemical anomalies and other information, aided by the history oil film thatinterpreted at the old times to determine contamination oil film and leakage oil film.