Spatial Convolution Grid Filters (T34)
Contents
1. USGS The aim here is to automatically classify zones in your geophysics grid that derive from distint underlying geology An example that makes immediate sense 1s to consider a Total count gamma ray grid and to look for regions with like counts and then making the jump to claim these regions define closed polygons of like geology The algorithm works by finding local maximum and minimum anomalies and working from these points to find the flexure lines defining the flexural change of gradient By iteratively filling the valleys and stripping the hills in the signal the so called terracing process bounding polygons emerge for each region INTREPID have attempted some further optimizations and extensions to this idea including writing out GIS polygons It turns out to be easily stated as above but hard in practise to get anything but the simplest of algorithms to work This is similar to a GIS Classifier algorithm Library Help Top 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 8 Library Help Top 4 Back gt Specifying a Tensor Grid Convolution Operation ml Intrepid Convolution Fitter File User Defined Kernels Preset Kernels Window Help Potential LaPlacian Error Fourier Coeffs Smoothed Tensor Estimate Vertical Component Two types of operations are supported Estimation of errors in the tensor grid via the use of a differ2. dataset you wish to enhance using the Spatial Convolution process Specify Output File Use this option to specify the name for the output grid dataset you are creating with the Spatial Convolution process Load Options If you wish to use an existing task specification file to specify the Spatial Convolution process use this option to specify the task specification file required INTREPID will load the file and use its contents to set all of the parameters for the Spatial Convolution process See Displaying options and using task specification files for more information Save Options If you wish to save the current Spatial Convolution Grid Filters file specifications and parameter settings as an task specification file use this option to specify the filename and save the file See Displaying options and using task specification files for more information User defined kernel files User defined kernel ker files reside in the install_path kernel directory When you display the User Defined Kernels menu INTREPID builds the menu from the names of files in that directory When you choose an item from the directory you load the corresponding user defined kernel file See User defined grid kernels supplied with INTREPID in INTREPID spatial and time domain filters and transformations R13 for information about the kernels supplied with INTREPID If you wish to add your own user defined kernels simply place them in this directory See C
3. INTREPID User Manual Spatial Convolution Grid Filters T34 1 Library Help Top q Back gt Spatial Convolution Grid Filters T34 Top Spatial Convolution is a method of enhancing a grid dataset The INTREPID Spatial Convolution Grid Filters tool compares the value of each grid cell with that of its neighbours and adjusts it using a formula based on the values of the cell itself and those of its neighbours The formulas that it uses are called Kernels There are three type of kernels User defined kernels are coefficient matrix filters which are defined in external ker files The files specify a weighting for the cell and its neighbours You can select the kernel that you require from the predefined kernels supplied with INTREPID or use a kernel you have created INTREPID displays an illustration of the kernel Preset kernels are statistical filters that apply a predefined algorithm to the cell and its neighbours These filters are not stored in external ker files Tensor grid filters that apply a predefined algorithm to the cell and its neighbours The basis function used here is a 3D spatial truncated Fourier series without the requirement that the samples be regularly spaced These filters are not stored in external ker files Using the Spatial Convolution Grid Filters tool Library Help Top gt gt To use Spatial Convolution Grid Filters with the INTREPID user interface 1 Choose Grid Convolution from the Filt
4. cify the window size and any other parameters required See Specifying a preset kernel for details If you select a user defined kernel specify the edge processing option and whether to normalise See Specifying a user defined kernel for details INTREPID displays a coloured pattern which graphically illustrates the user defined kernel a CS File User Defined Kernels Preset Kernels Tensor Convolutions Window Help Chop at edge Normalise weights l Save Residual Apply 6 Specify whether you require the output grid dataset to contain the convolved corrected values or values of the corrections only If you wish to save the corrections only turn on the Save Residuals box in the Spatial Convolution Grid Filters main window See Apply for instructions 7 When you have made specifications and settings according to your requirements choose Apply INTREPID will perform the Spatial Convolution process and save the results to the specified output dataset 8 If you wish to record the specifications for this process in a task specification job file in order to repeat a similar task later or for some other reason use Save Options from the File menu See Specifying input and output files for detailed instructions 9 If you wish to repeat the process repeat steps 2 8 varying the kernels parameters and or data files as required 10 To exit from Spatial Convolution Grid Filters choose Quit from the File menu To vie
5. el files have the extension ker Next INTREPID displays the Kernel Filter Attributes dialog box Kernel Filter Attributes C Chop Kernel C Null Kernel Normalise _OK Cancel 3 Specify the parameters then choose OK Library Help Top 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 5 Library Help Top 4 Back gt Kernel filter attributes Chop kernel Null kernel When INTREPID is processing the edge of the grid the convolution formula may require values from cells that do not exist because they would be outside the grid You can use the Chop Kernel Null Kernel option buttons to specify the action that INTREPID should take at the edge of the grid Chop Kernel If you choose this option INTREPID will reduce the number of cells used in the kernel at the edges of the grid so that it does not attempt to refer to nonexistent cells Null Kernel If you choose this option INTREPID will allocate null to a target cell if a nonexistent cell is within the scope of the kernel This is likely to happen along dataset edges or near data gaps Normalising kernels If you turn on the Normalise check box INTREPID will adjust the kernel so that the sum of its weights is one The output grid will then have the same mean value as the input grid Creating and editing user defined convolution kernel files You can use any text editor to create and edit convolution kern
6. els You can define your own from scratch or use a supplied kernel file as a template If you define further kernels and store them in install_path kernel with extension ker INTREPID will automatically include them in the User Defined Kernels menu See User defined convolution kernel ker files in INTREPID spatial and time domain filters and transformations R138 for further details User defined kernel display INTREPID displays an illustration of the kernel that you have selected in a display box in the main window The box contains a matrix of squares each assigned a colour The central square represents the target cell of the formula The other squares represent the neighbouring cells that are involved in the formula The colour of each square represents the weight that it carries in the kernel formula Red cells carry the highest weight and blue cells carry the lowest weight A green cell carries an average weight usually 0 Many of the convolution kernels naturally have the target cell carrying the highest weight In these cases the Kernel Illustration has one red square and a number of ereen and blue squares Library Help Top 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 6 Library Help Top lt 4 Back gt The number of squares and colours in a kernel illustration box varies according to the number of cells involved in the formula and the number of di
7. ence of the estimated tensor from neighbouring cells to what is contained at the current cell Estimation of the Local Potential smoothing this and then re estimating a smoothed tensor grid A Fourier Series is fitted to the tensor observations in a local 3 5 3 samples row pattern The coefficients are solved for using a Single Value Decomposition so as to get a best fit in a least squares sense You can save off the 48 coefficients to a multi band grid If this coefficients grid and the potential grid are the basis for being able to estimate any tensor or vector component or gradient at every point in the grid It is surprising how similar the Potential grid can look to the vertical component of gravity This is mostly a mathematical exercise and it enables a quite viable grid smoothing technique that contrasts with the MITRE FTG technique The final option is the estimation of the vertical component of gravity from this set of coefficients and the potential As this is derived from a local 5 5 window it has a lot of the character of a high passed version of the vertical component Convolution status and progress reporting Kernel description rows columns INTREPID displays information in the status area at the top of the Command area of the Spatial Convolution Grid Filters Main window INTREPID uses status area to display a brief description of the current kernel and the number of rows and columns of neighbouring cells that are invo
8. ering menu in the Project Manager or type the command cfilter exe INTREPID displays the Spatial Convolution Grid Filters Main window z Ey Intrepid Convolution Fite MMM File User Defined Kernels Preset Kernels Tensor Convolutions Window Help l Save Residual Apply 2 Ifyou have previously prepared file specifications and parameter settings for Spatial Convolution Grid Filters load the corresponding task specification file using Load Options from the File menu See Specifying input and output files for detailed instructions If all of the specifications are correct in this file go to step 7 If you wish to modify any settings carry out these steps as required 3 Specify the grid dataset on which you wish to perform the Spatial Convolution Choose Open Input Dataset from the File menu See Specifying input and output files for detailed instructions 4 Specify the output grid dataset to be created with the results of the process 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 2 Library Help Top Library Help Top lt 4 Back gt Choose Specify Output File from the File menu See Specifying input and output files for detailed instructions 5 Specify the convolution filter you wish to use from the User Defined Kernels menu Preset Kernels or Tensor menu If you select a statistical filter from the Preset Kernels menu spe
9. fferent weights assigned to cells INTREPID automatically selects the numbers of squares and colours to provide the optimum illustration of each kernel a FI Intrepid Convolution Fite e File User Defined Kernels Preset Kernels Tensor Convolutions Window Help Gauss5x_02 ker Rows 5 Cols 5 Chop at edge Normalise weights Save Residual Apply Specifying a preset kernel In addition to the kernels provided in ker files the Spatial Convolution Grid Filters tool has a number of internal filters which apply a predefined algorithm to the cell and its neighbours You can specify these using options from the Preset Kernels menu Fl Intrepid Convolution Fil File User Defined Kernels Tensor Convolutions Wine Median CNorm Gaussian Anti alias Terrace Local median Filter Library Help Top This filter adjusts the value of the target cell based on the median value of the cell and its neighbours in a window To use the local median filter Choose Median from the Preset Kernels menu INTREPID displays the Local Median Filter for Grid dialog box Median Filter Box Window 21 OK Cancel Specify the window size then choose OK 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 7 Library Help Top 4 Back gt See Local mean median filters in INTREPID spatial and time domain filters and transformations R13 for fu
10. ile called surv329 job in the current directory you would use the command cfilter exe batch surv329 job Library Help Top 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 11 Library Help Top 4 Back gt Task specification file notes and example Here is an example of a Spatial Convolution Grid Filters task specification file Process Begin Name Convolve InputImage disk1 data mlevel_grid OutputImage disk1 data mlevel_C Parameters Begin Type Convolve Kernel Begin Kernel low_pass_ 4 ker NormaliseWeights Yes ChopWeights Yes NullWeights No Kernel End Residual No Parameters End Process End Frequently asked questions Q Can I run an AGC filter on a grid The CNorm Spatial Contrast Normalisation filter is a non linear space variant stretch filter which enhances the low amplitude high frequency content and dampens the high amplitude content It operates on gridded data and has a similar effect to an Automatic Gain Control filter It is located under Preset Kernels gt CNorm Library Help Top 2012 Intrepid Geophysics 4 Back gt
11. ing options To display the current file specifications and parameter settings choose Report from the Window menu File User Defined Kernels Preset Kernels Help Main Report INTREPID displays the Spatial Convolution Grid Filters Report window containing the task specifications Using task specification files You can store sets of file specifications and parameter settings for Spatial Convolution Grid Filters in task specification job files gt gt To create a task specification file with the Spatial Convolution Grid Filters tool 1 Specify all files and parameters 2 If possible execute the task choose Apply to ensure that it will work 3 Choose Save Options from the File menu Specify a task specification file INTREPID will add the extension job INTREPID will create the file with the current parameters For full instructions on creating and editing task specification files see INTREPID task specification job files R06 gt gt To use a task specification file in an interactive Spatial Convolution Grid Filters session Load the task specification job file File menu Load Options modify any settings as required then choose Apply gt gt To use a task specification file for a batch mode Spatial Convolution Grid Filters task Type the command cfilter exe with the switch batch followed by the name and path if necessary of the task specification file For example if you had a task specification f
12. lved in the formula Library Help Top 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Library Help Top Apply Spatial Convolution Grid Filters T34 9 lt 4 Back gt After you have specified the spatial convolution filter required you can apply the filter to the grid gt gt To apply the filter 1 Use the Save Residual check box in the command area to specify whether to save 1 New filtered grid turn off or 2 Values of changes to the grid turn on 2 Choose Apply INTREPID will apply the filter and save the results Help You can use the Help menu to display help text on the topics shown in the menu illustration below P Intrepid Convolution Fite Sh SaR File User Defined Kernels Preset Kernels Tensor Convolutions Window Help Spatial Convolution PDF INTREPID Contents PDF INTREPID Search PDF Website Library Main Window gt File User Defined Kernels oe el aa es so Et re Preset Kernels Window Terrace Gauss Tensor Task Specification Files About prep Tr Spatial Convolution Grid Filters To exit from Spatial Convolution Grid Filters choose Quit from the File menu Library Help Top 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 10 Library Help Top 4 Back gt Displaying options and using task specification files Display
13. reating and editing user defined convolution kernel files for instructions Library Help Top 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 4 Library Help Top 4 Back gt Specifying a user defined kernel You can select the kernel that you require from the predefined kernels supplied with INTREPID You can also edit the kernels supplied or create your own kernel gt gt To select an INTREPID predefined Convolution kernel 1 Under the User Defined Kernels menu choose Select From List P Intrepid Convolution Filter Il Sclect Convolve Kernel EE File User Defined Kernels Preset Kernel 3x3ave ker a Select F Li 3x3tri ker settee 5x5ave ker 5x5tri ker x ave ker x triker 9x9ave ker 9x9tri ker East_West ker Gauss3x_0d6 ker Gauss5x_02 ker North_East ker North_South ker North_ West ker bandpass _3_5 ker down_cont_1 ker down_cont_2 ker down_cont_half ker downhalf ker firstvertical ker h_edge ker h_edge_5 ker high_pass_4 ker hz10_2nd_deriv ker hz13_2nd_deriv ker hz1960_2nd_deriv ker low_pass_4 ker nsgrad ker residual_1 _ker residual_2_ ker secondvertical ker sharpedge ker i up_cont_1 ker up_cont_2 ker nomitre_16m isi up_cont_half ker wiener ker 2 From the Select Kernel list choose the filter option you wish to use The Select Kernel list lists all files in the install_path kernel directory The kern
14. rther instructions Contrast Normalisation Filter This is a non linear space variant stretch filter that will enhance the low amplitude high frequency content and dampen the high amplitude content It has a similar effect to an AGC Automatic Gain Control filter gt gt To use a contrast normalisation filter Choose CNorm from the Preset Kernels menu See Contrast normalisation filter for grid data in INTREPID spatial and time domain filters and transformations R18 for further explanation of parameters Gaussian Anti alias filter This is a standard convolution filter whose aim is to look for non Gaussian aspects to the signal in any one direction within the window and normalise the grid values by nudging The argument for this style of filter arises when you have much higher frequency data along survey profile lines as you may have a sample every 6 meters than across the lines as the line seperation maybe say 100m with a 25m cellsize Often a low pass filter such as a Butterworth can be applied along the profile to more or less match the along profile wavelengths to what can be successfully captured in the across line direction In the case mentioned above wavelengths less than 100m if left in the profiles while gridding will cause frequency aliasing to occur in the grid This Gaussian spatial convolution filter can then be applied to try and remedy the situation Terracing filter This is an interesting idea from the
15. w the current set of specifications choose Report from the Window menu INTREPID displays the Spatial Convolution Grid Filters Report window See Displaying options and using task specification files for details and an example of a set of specifications You can view Help information by choosing options from the Help menu See Help You can execute Spatial Convolution Grid Filters in batch mode using a task 2012 Intrepid Geophysics 4 Back gt INTREPID User Manual Spatial Convolution Grid Filters T34 3 Library Help Top lt 4 Back gt specification job file that you have previously prepared See Displaying options and using task specification files for details Specifying input and output files To use Spatial Convolution Grid Filters you will need to specify the name of the grid dataset you wish to process and a name for the output grid dataset containing the results Choose the options as required from the File menu E Intrepid Convolution Filter W File User Defined Kernels Preset Kernels Open Input Grid Dataset Specify Output File Load Options Save Options Quit In each case INTREPID displays an Open or Save As dialog box Use the directory and file selector to locate the file you require See Specifying input and output files in Introduction to INTREPID R02 for information about specifying files File menu options Open Input Dataset Use this option to specify a grid
Download Pdf Manuals
Related Search