Font Size: a A A

Study On Differential Box-counting And Its Application Of Image Segmentation

Posted on:2016-11-17Degree:MasterType:Thesis
Country:ChinaCandidate:L F ChengFull Text:PDF
GTID:2308330479450182Subject:Instrumentation engineering
Abstract/Summary:PDF Full Text Request
In the industrial production field industrial production field,lighting conditions and nonlinear characteristics of part itself lead to image mutation of the detection unit.in this case,the changed industrial parts image have to be segmented to measure the geometric parameters of industrial parts.Many image segmentation methods emerged,but each method has different advantages and disadvantages.In recent decades, fractal geometry theory has solved a big difficulty that traditional geometry could not descript and process geometry with a strongly irregular edge.the establishment of fractal theory opens a new route for image segmentation technology.Relative to the traditional image segmentation algorithm,method based on fractal geometry can detect almost all of the image details,and also has better noise resistance.Therefore, the research of this paper has important engineering significance.specific work is as follows:(1)A image segmentation method based on differential box-counting dimension is roposed.Differential box-counting dimension method is used to calculate the the fractal dimension of each area in the original image.The fractal dimension is used to construct a three-dimensional map. The three-dimensional map is combined with Ostu algorithm for image segmentation.(2)The improvement of differential box-counting dimension. The improvement is from the two aspects of linear fitting precision and computing speed.Linear fitting is improved through calculating the grey scale within each child window and extending the size of child window.So that,the fitting error associated with calculating the fractal dimension is reduced.The speed is improved through preserving maximum and minimum grey value of each size for after using.So that,the execution time of the algorithm. Is mproved.(3)Screw image segmentation based on improved differential box-counting method.based on improved differential box-counting method and Ostu algorithm are combined to segment screw image and screw image without grain.The result of experiments show that compared with the traditional differential box-counting method is more precise when calculating the fractal dimension.,better expressing the texture structure of image surface,and reducing operation time.
Keywords/Search Tags:fractal geometry, differential box-counting dimension, Ostu, image segmentation
PDF Full Text Request
Related items