Application Of Color Image Processing In Industrial Parameters Detection

Taking the dual-coloured water-level-gauges as object, this thesis introduces some knowledge about industrial color image processing,and detects the drum water level automatically.In the fieldwork of industry, many disadvantages will cause that the obtained images are fuzzy, noisy or will bring characters into scaling and shift. Based on spatial domain processing and BP network, what the author has studied and realized mainly are as follows: (1) Presenting a novel color image filtering method based on pixels type, which has effective filtering performance for various noise; (2) Improving the mini-objects segmentation mothod based on 2D Otsu thresholding; (3)Utilizeing BP network based on fuzzy features as instrument digits'recognizer; (4) Using BP network to fit the H-curve in HSI color space and realizing the estimation of water level. All the algorithms above are experimentally validated and analyzed.