The Research Of Classification And Recognition Of The Liquid Level System Image In The Complex Industrial Scene

The thesis studied the classification and recognition of the liquid level system imagein the complex industrial scene. Taking the pattern recognition process as the main line,the author took the method of combining the theoretical analysis with simulationexperiment, and carried on the design on the basis of image acquisition, imagepreprocessing, image segmentation, feature extraction and pattern classification to realizethe purpose.On the stage of image preprocessing, based on the classical median filtering, animproved multilevel weighted mean filter algorithm with filter window was proposed andit had better filtering effect on the liquid level system image in the complex industrialscene under the condition of spiced salt and pepper noise and Gaussian noise. On the stageof image segmentation, the author proposed an improved algorithm that fused the grayhistogram peaks and K-means clustering algorithm, which had a more complete andaccurate image segmentation effect. On the stage of image feature extraction, a kind ofimage integrated features extraction method based on shape features, color features andtexture features was put forward, which successfully extracted the main characteristics ofthe image. And on the stage of image classification, a kind of strong weights updated BPclassifier algorithm that classified two kinds of image in level system images in thecomplex industrial scene was successfully applied to.Through in-depth study on classification and recognition of the liquid level systemimages in the complex industrial scene, this thesis laid a theoretical foundation for therealization of the industrial scene recognition and the mobile robot in the industrial siteoperation. Appling the mobile robot which had the ability of image classificationrecognition into the industrial field, such as thermal power plants, power plants, etc, wherehad high temperature, high radiation, and high pollution and was bad for workers to workin a long time, to a great extent, would realize unmanned and automatic operation incomplex industrial process, which had practical application value.