I have used a simple algo although it is not very efficient but it will fulfills my requirements. ![]() I have used the following approach may be u can help me in my algo If u can please tell me that how have u detected the place where the number plate is kindly tell me the algo The second one is to identify the place where the license plate is using aForge.Net blobs recognition. Removal of noise can be achieved by a convulsion of a mask to remove 'salt and pepper' type noise. One consideration could be the orientation of the the image is the plate at 90 degrees to your sample? If not you could have the symbol 'A' and an image of that symbol but because they are at different orientations the mean squared error would suggest there wasn't a match. ![]() Locating the licence plate however would be very difficult, similarly if we assume you're not using a perfect sample, pre-processing would take some thought. The artefacts you then retrieve could be compared to known symbols, using some metric such as mean squared error you could determine the most likely match. Aside from locating the licence plate in an image, if you had a perfect noise free example, with no erroneous artefacts, edge detection algorithms would be best suited for your analysis. Unfortunately this is not a simple operation as advanced image analysis is not a trivial area of research.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |