Study On Automatic Scaling Of Ionospheric Vertical Sounding Ionogram Based On Image Processing

As the "mirror of solar activity" and "magnifier of atmospheric disturbance",ionospheric research has a very important academic significance and socio-economicvalue. Ionospheric characteristic parameters can reflect the ionospheric status,however, ionospheric characteristic parameters are mainly obtained from manuallyscaling the ionogram by the well-experienced operators, which is not onlytime-consuming and laborious, but can not avoid the scaling error caused by thesubjective factors. Consequently the automatic scaling of the ionograms has beensubject to extensive attention from researchers.The paper is based on the TYC-1ionosondes developed by China ResearchInstitute of Radio Wave Propagation in the observatories of different latitudes and theobjects of study are the conventional vertical sounding ionograms recorded in thecalm state of the ionosphere. By analyzing the theory of the characteristics ofionospheric structure, vertical sounding ionograms and the measure of ionosphericcharacteristic parameters, learning from the automatic scaling methods home andabroad, image processing technology is the main means to research and explore themethod about autoscaling of ionogram, though “ionograms pretreatment-ionogramssegmentation-ionograms identification". The main works of this paper are as follows:1. In the ionograms pretreatment, first convert the sounding data to gray-levelimage based on the detection principle of the ionosonde; second filter out noise andget the binary ionogram image, combined with Gray-scale morphological operation,neighborhood enhance and the Otsu method according to the characteristics ofionospheric vertical echoes. Meanwhile the shape of the trace is preserved.And thecontrast experiments show the effectiveness of the pretreatment method.2. In the ionograms segmentation, in view of strong interference in high frequencyarea, extract useful channels by the image vertical projection method based on theregion-wide of the trace; seperate E region and F region using a dividing line localization algorithm combined with the characteristics of ionospheric structureand the height range of E-F valley. The experiments prove that the algorithm isfeasible and effective.3. In the ionograms identification, mainly use image processing technology to（1） identify the trace, scale parameters of F region.（2） detect the trace in Eregion.（1） Use connected components label, image rotation and projection to removethe second-order F layer reflections combined with the similarity and regionallocation relations between them; use Hough transform to detect the F critical traceand connect the fracture; for the F O-ray traces and X-ray traces overlapping, firstuse image skeleton, skeleton critical point detection and skeleton decomposition toextract the skeleton of O-ray traces, then extract the O-ray traces combined withmorphological operations; for the stratification of F layer, use the polynomial leastsquares fitting method to determine the hierarchical point of F1and F2layer; get theF-layer parameters use the image projection method based on ionograms measurerules. Experiments show that this method can achieve good effect withoutconsidering the blanket and interference of multiple reflections from Es layer.（2） Based on ionograms measure rules, the statistical analysis of the tracecharacteristics and the height range of single reflection trace in E region, useconnected components label and image projection method to detect the trace in Eregion in order to prepare for the measure of E-layer and Es-layer parameters whileno trace exists. The experiments achieve good results and verify the feasibility ofthis method.