| Under wants to divide two parts to give the briefing to the full text content.The graph boundary localization method image majority of informations all exist in the image edge, main performance for image partial characteristic discontinuity, namely in image gradation change quite fierce place.Therefore we define the edge for the image in the gradation occur the fierce change region boundary. Edge examination step:(1) The image filter(2) image intensification(3) image examination(4) image locatesThe edge examination algorithm:(1) Traditional edge examination operator.Including based on gradation histogram edge examination, based on gradient edge examination, Laplacan edge operator and so on.(2) Canny edge examination operator.(3) Fuzzy reasoning edge examination.(4) Mallat wavelet edge examination operator. The author has designed a set of testing plan, after has contrasted the wavelet algorithm and the traditional operator processing graph boundary.Tests the concrete step to be as follows:(1). lasers send out the light microscope optics imaging system becomes the circular light spot .CCD real-time recording facula image on the receive screen, transforms two value digitization image after technical processing, andspreads to the computer storage.Stores up the image after the imagery processing system filter, decides the boundary, obtained empirical datum and image display on computer. (2) Causes light beam relatively initial position towards left (right) which the laser sends out precisely to move 30,60,90,120um in turn ......Under, and records separately with CCD receives on the screen the facula image, moves 9 times, altogether 10 images.(3) Separately results in the boundary image to above 10 image application tradition boundary algorithm and wavelet boundary algorithm processing.No matter what takes in the image some line of data, contrast tradition boundary algorithm and wavelet boundary algorithm effect. The experimental contrast result is as follows: In the synthesis, the author thought that,The wavelet algorithm is one kind of more precise boundary localization algorithm. The graph area examination method regarding the area survey, the commonly used method has the following several kinds: Mathematical analysis law; Each kind of area metering equipment; Specialized examination software and so on. The mathematical analysis law computation load is big, difficulty with considered, and lives many image not obvious boundaries, or the boundary too is complex, therefore is simpler regarding the stochastic random graph area survey need the intelligent method. The measuring instrument measuring accuracy decided by the obvious satellite number that, the area usually measures by the square meter or the Chinese acre, because the precision is not high, therefore usually applies in the farmland, green space, forest, waters constant area survey. Some specialized softwares may carry on the examination to the graph area.If CAD, Fotoshop and so on.Although this method simple, but the measuring accuracy is fixed, and is not high. This article in this proposed the application digital image processing method examination graph area, and designs the experimental actual survey. The experimental concrete step is as follows:(1) In front of CCD fixed spotting board, treats measures the graph smooth and level (or image formation) on the spotting board (the spotting board As soon as passes through fixedly, sends the experiment to finish completely, the overall system does not have to move).(2) The recording image of CCD, transforms two value digitization image after technical processing, and spreads to the computer storage.Stores up the image after the image pretreatment system filter, after decides the boundary the obtained empirical datum and the image display on the computer.(3) Selects certain known area the graph to do treats measures the graph, after the system computation, obtains the picture element integer which the graph image holds.(4) Selects the graph willfully, the picture element integer which the computation graph image holds, the substitution specific formula operation results in the graph area.In to the static graph area survey in, the error is approximatelyσ=0.0970 cm 2. The analysis source of error is as follows:(1) Uses in the demarcation pan which demarcates is not is vertical completely in tests the platform, creates the photography image to have the distortion slightly.(2) The camera photography effect is not very ideal, the boundary is not clear.(3) Uses in scanning the image picture element the scanning procedure not to be able to process the good image boundary to have in half picture element situation the solution.This also is one of this system pending solution questions.To the handpainted dynamic graph examination, the author also proposed own opinion, elaborated the principle and the concrete experimental technique in detail.Because the earlier period has used the wavelet boundary localization method, the light pen pointing accuracy achieved 10um . Conclusion: This article applies the wavelet algorithm in the imagery processing, mainly discussed the graph boundary localization as well as the graph area examination essential method.Indicated through the experiment that, after wavelet technology processing, the graph boundary localization precision large scale enhancement, achieved um .The boundary localization is the area examination premise, along with the boundary pinpointing, the graph area examination system survey relative error also limited in 4/1000 scopes.Although this examination system precision is very high, but still had the improvement leeway.In the graph boundary localization application, through to the wavelet theory thorough understanding, selects the more appropriate wavelet primary function regarding to increase the boundary pointing accuracy to have the very big help.In graph area examination application, if can solve the photography process Chinese Library image distortion problem, arranges the imaging system reasonably, causes the CCD image formation to be clearer, then this system precision also can enhance large scale. |
PDF Full Text Request