The Study On Image Preprocessing And Image Segmentation Of Image Liquid Level Detection

The traditional level detection methods have various shortcomings.For examples,the ultrasonic type is easily to be influenced by the fluctuations and it is veryexpensive; the capacitive type is only suitable for the detection of the liquid that havelarge dielectric constant. So these traditional level detection methods are difficultto obtain satisfactory results in these situations that surrounding environment iscomplex or high measurement accuracy is requiremented. So, there has been a lot ofnew detection methods,in these liquid level detection methods, the image processingmethod has a high widespread praise for its advantages such as high accuracy,good real-time, non-contact and good adaptability to environment.The image level detection is a very complex process, the image preprocessingand image segmentation are two essential parts of the image level detection, and arethe basis of the follow-up to level line extraction, are the key steps in the detectionprocess. Only realize the accurate segmentation of the level image that it is possible toensure the subsequent extraction of the liquid level line height with good accuracy,and accurate image segmentation need the image preprocessing to provide thenecessary conditions, so it is very necessary to do some research on imagepreprocessing and image segmentation in the process of of liquid level detection. Thecompleted work of this paper is as follows:(1) In the aspect of image preprocessing:introduced many image grayscaleenhancement algorithms,adopted the logarithmic method to realize the grayscaleenhancement; did some comparative analysis on image filtering algorithms includeneighborhood average, median filter and Wiener filter methods, analyzed theiradvantages and disadvantages combined with the actual level pictures，chose themedian filtering method which had a better effect to level image filtering de-noising.The experimental results show that the proposed algorithm can achieve satisfactoryresults.(2) In the aspect of image segmentation: elaborated various image segmentationalgorithms, made a focus on the description of the image segmentation method basedon edge detection including Sobel operator, Prewitt operator, Canny operator, etc, theexperimental results showed that the Canny operator had a better effect, and animproved Canny edge detection algorithm based on self-adaptive threshold wasproposed. The experimental results of the image show that the improved Canny operator overcomed the shortcomings of traditional Canny operator which lose weakedges and increase false edges more easy and the new operator had a moreaccurate segmentation result.