diff options
Diffstat (limited to 'html/jpgraph/jpgraph_contour.php')
| -rw-r--r-- | html/jpgraph/jpgraph_contour.php | 587 |
1 files changed, 587 insertions, 0 deletions
diff --git a/html/jpgraph/jpgraph_contour.php b/html/jpgraph/jpgraph_contour.php new file mode 100644 index 0000000..760989e --- /dev/null +++ b/html/jpgraph/jpgraph_contour.php @@ -0,0 +1,587 @@ +<?php +/*======================================================================= +// File: JPGRAPH_CONTOUR.PHP +// Description: Contour plot +// Created: 2009-03-08 +// Ver: $Id: jpgraph_contour.php 1870 2009-09-29 04:24:18Z ljp $ +// +// Copyright (c) Asial Corporation. All rights reserved. +//======================================================================== +*/ +require_once('jpgraph_meshinterpolate.inc.php'); +define('HORIZ_EDGE',0); +define('VERT_EDGE',1); + +/** + * This class encapsulates the core contour plot algorithm. It will find the path + * of the specified isobars in the data matrix specified. It is assumed that the + * data matrix models an equspaced X-Y mesh of datavalues corresponding to the Z + * values. + * + */ +class Contour { + + private $dataPoints = array(); + private $nbrCols=0,$nbrRows=0; + private $horizEdges = array(), $vertEdges=array(); + private $isobarValues = array(); + private $stack = null; + private $isobarCoord = array(); + private $nbrIsobars = 10, $isobarColors = array(); + private $invert = true; + private $highcontrast = false, $highcontrastbw = false; + + /** + * Create a new contour level "algorithm machine". + * @param $aMatrix The values to find the contour from + * @param $aIsobars Mixed. If integer it determines the number of isobars to be used. The levels are determined + * automatically as equdistance between the min and max value of the matrice. + * If $aIsobars is an array then this is interpretated as an array of values to be used as isobars in the + * contour plot. + * @return an instance of the contour algorithm + */ + function __construct($aMatrix,$aIsobars=10, $aColors=null) { + + $this->nbrRows = count($aMatrix); + $this->nbrCols = count($aMatrix[0]); + $this->dataPoints = $aMatrix; + + if( is_array($aIsobars) ) { + // use the isobar values supplied + $this->nbrIsobars = count($aIsobars); + $this->isobarValues = $aIsobars; + } + else { + // Determine the isobar values automatically + $this->nbrIsobars = $aIsobars; + list($min,$max) = $this->getMinMaxVal(); + $stepSize = ($max-$min) / $aIsobars ; + $isobar = $min+$stepSize/2; + for ($i = 0; $i < $aIsobars; $i++) { + $this->isobarValues[$i] = $isobar; + $isobar += $stepSize; + } + } + + if( $aColors !== null && count($aColors) > 0 ) { + + if( !is_array($aColors) ) { + JpGraphError::RaiseL(28001); + //'Third argument to Contour must be an array of colors.' + } + + if( count($aColors) != count($this->isobarValues) ) { + JpGraphError::RaiseL(28002); + //'Number of colors must equal the number of isobar lines specified'; + } + + $this->isobarColors = $aColors; + } + } + + /** + * Flip the plot around the Y-coordinate. This has the same affect as flipping the input + * data matrice + * + * @param $aFlg If true the the vertice in input data matrice position (0,0) corresponds to the top left + * corner of teh plot otherwise it will correspond to the bottom left corner (a horizontal flip) + */ + function SetInvert($aFlg=true) { + $this->invert = $aFlg; + } + + /** + * Find the min and max values in the data matrice + * + * @return array(min_value,max_value) + */ + function getMinMaxVal() { + $min = $this->dataPoints[0][0]; + $max = $this->dataPoints[0][0]; + for ($i = 0; $i < $this->nbrRows; $i++) { + if( ($mi=min($this->dataPoints[$i])) < $min ) $min = $mi; + if( ($ma=max($this->dataPoints[$i])) > $max ) $max = $ma; + } + return array($min,$max); + } + + /** + * Reset the two matrices that keeps track on where the isobars crosses the + * horizontal and vertical edges + */ + function resetEdgeMatrices() { + for ($k = 0; $k < 2; $k++) { + for ($i = 0; $i <= $this->nbrRows; $i++) { + for ($j = 0; $j <= $this->nbrCols; $j++) { + $this->edges[$k][$i][$j] = false; + } + } + } + } + + /** + * Determine if the specified isobar crosses the horizontal edge specified by its row and column + * + * @param $aRow Row index of edge to be checked + * @param $aCol Col index of edge to be checked + * @param $aIsobar Isobar value + * @return true if the isobar is crossing this edge + */ + function isobarHCrossing($aRow,$aCol,$aIsobar) { + + if( $aCol >= $this->nbrCols-1 ) { + JpGraphError::RaiseL(28003,$aCol); + //'ContourPlot Internal Error: isobarHCrossing: Coloumn index too large (%d)' + } + if( $aRow >= $this->nbrRows ) { + JpGraphError::RaiseL(28004,$aRow); + //'ContourPlot Internal Error: isobarHCrossing: Row index too large (%d)' + } + + $v1 = $this->dataPoints[$aRow][$aCol]; + $v2 = $this->dataPoints[$aRow][$aCol+1]; + + return ($aIsobar-$v1)*($aIsobar-$v2) < 0 ; + + } + + /** + * Determine if the specified isobar crosses the vertical edge specified by its row and column + * + * @param $aRow Row index of edge to be checked + * @param $aCol Col index of edge to be checked + * @param $aIsobar Isobar value + * @return true if the isobar is crossing this edge + */ + function isobarVCrossing($aRow,$aCol,$aIsobar) { + + if( $aRow >= $this->nbrRows-1) { + JpGraphError::RaiseL(28005,$aRow); + //'isobarVCrossing: Row index too large + } + if( $aCol >= $this->nbrCols ) { + JpGraphError::RaiseL(28006,$aCol); + //'isobarVCrossing: Col index too large + } + + $v1 = $this->dataPoints[$aRow][$aCol]; + $v2 = $this->dataPoints[$aRow+1][$aCol]; + + return ($aIsobar-$v1)*($aIsobar-$v2) < 0 ; + + } + + /** + * Determine all edges, horizontal and vertical that the specified isobar crosses. The crossings + * are recorded in the two edge matrices. + * + * @param $aIsobar The value of the isobar to be checked + */ + function determineIsobarEdgeCrossings($aIsobar) { + + $ib = $this->isobarValues[$aIsobar]; + + for ($i = 0; $i < $this->nbrRows-1; $i++) { + for ($j = 0; $j < $this->nbrCols-1; $j++) { + $this->edges[HORIZ_EDGE][$i][$j] = $this->isobarHCrossing($i,$j,$ib); + $this->edges[VERT_EDGE][$i][$j] = $this->isobarVCrossing($i,$j,$ib); + } + } + + // We now have the bottom and rightmost edges unsearched + for ($i = 0; $i < $this->nbrRows-1; $i++) { + $this->edges[VERT_EDGE][$i][$j] = $this->isobarVCrossing($i,$this->nbrCols-1,$ib); + } + for ($j = 0; $j < $this->nbrCols-1; $j++) { + $this->edges[HORIZ_EDGE][$i][$j] = $this->isobarHCrossing($this->nbrRows-1,$j,$ib); + } + + } + + /** + * Return the normalized coordinates for the crossing of the specified edge with the specified + * isobar- The crossing is simpy detrmined with a linear interpolation between the two vertices + * on each side of the edge and the value of the isobar + * + * @param $aRow Row of edge + * @param $aCol Column of edge + * @param $aEdgeDir Determine if this is a horizontal or vertical edge + * @param $ib The isobar value + * @return unknown_type + */ + function getCrossingCoord($aRow,$aCol,$aEdgeDir,$aIsobarVal) { + + // In order to avoid numerical problem when two vertices are very close + // we have to check and avoid dividing by close to zero denumerator. + if( $aEdgeDir == HORIZ_EDGE ) { + $d = abs($this->dataPoints[$aRow][$aCol] - $this->dataPoints[$aRow][$aCol+1]); + if( $d > 0.001 ) { + $xcoord = $aCol + abs($aIsobarVal - $this->dataPoints[$aRow][$aCol]) / $d; + } + else { + $xcoord = $aCol; + } + $ycoord = $aRow; + } + else { + $d = abs($this->dataPoints[$aRow][$aCol] - $this->dataPoints[$aRow+1][$aCol]); + if( $d > 0.001 ) { + $ycoord = $aRow + abs($aIsobarVal - $this->dataPoints[$aRow][$aCol]) / $d; + } + else { + $ycoord = $aRow; + } + $xcoord = $aCol; + } + if( $this->invert ) { + $ycoord = $this->nbrRows-1 - $ycoord; + } + return array($xcoord,$ycoord); + + } + + /** + * In order to avoid all kinds of unpleasent extra checks and complex boundary + * controls for the degenerated case where the contour levels exactly crosses + * one of the vertices we add a very small delta (0.1%) to the data point value. + * This has no visible affect but it makes the code sooooo much cleaner. + * + */ + function adjustDataPointValues() { + + $ni = count($this->isobarValues); + for ($k = 0; $k < $ni; $k++) { + $ib = $this->isobarValues[$k]; + for ($row = 0 ; $row < $this->nbrRows-1; ++$row) { + for ($col = 0 ; $col < $this->nbrCols-1; ++$col ) { + if( abs($this->dataPoints[$row][$col] - $ib) < 0.0001 ) { + $this->dataPoints[$row][$col] += $this->dataPoints[$row][$col]*0.001; + } + } + } + } + + } + + /** + * @param $aFlg + * @param $aBW + * @return unknown_type + */ + function UseHighContrastColor($aFlg=true,$aBW=false) { + $this->highcontrast = $aFlg; + $this->highcontrastbw = $aBW; + } + + /** + * Calculate suitable colors for each defined isobar + * + */ + function CalculateColors() { + if ( $this->highcontrast ) { + if ( $this->highcontrastbw ) { + for ($ib = 0; $ib < $this->nbrIsobars; $ib++) { + $this->isobarColors[$ib] = 'black'; + } + } + else { + // Use only blue/red scale + $step = round(255/($this->nbrIsobars-1)); + for ($ib = 0; $ib < $this->nbrIsobars; $ib++) { + $this->isobarColors[$ib] = array($ib*$step, 50, 255-$ib*$step); + } + } + } + else { + $n = $this->nbrIsobars; + $v = 0; $step = 1 / ($this->nbrIsobars-1); + for ($ib = 0; $ib < $this->nbrIsobars; $ib++) { + $this->isobarColors[$ib] = RGB::GetSpectrum($v); + $v += $step; + } + } + } + + /** + * This is where the main work is done. For each isobar the crossing of the edges are determined + * and then each cell is analyzed to find the 0, 2 or 4 crossings. Then the normalized coordinate + * for the crossings are determined and pushed on to the isobar stack. When the method is finished + * the $isobarCoord will hold one arrayfor each isobar where all the line segments that makes + * up the contour plot are stored. + * + * @return array( $isobarCoord, $isobarValues, $isobarColors ) + */ + function getIsobars() { + + $this->adjustDataPointValues(); + + for ($isobar = 0; $isobar < $this->nbrIsobars; $isobar++) { + + $ib = $this->isobarValues[$isobar]; + $this->resetEdgeMatrices(); + $this->determineIsobarEdgeCrossings($isobar); + $this->isobarCoord[$isobar] = array(); + + $ncoord = 0; + + for ($row = 0 ; $row < $this->nbrRows-1; ++$row) { + for ($col = 0 ; $col < $this->nbrCols-1; ++$col ) { + + // Find out how many crossings around the edges + $n = 0; + if ( $this->edges[HORIZ_EDGE][$row][$col] ) $neigh[$n++] = array($row, $col, HORIZ_EDGE); + if ( $this->edges[HORIZ_EDGE][$row+1][$col] ) $neigh[$n++] = array($row+1,$col, HORIZ_EDGE); + if ( $this->edges[VERT_EDGE][$row][$col] ) $neigh[$n++] = array($row, $col, VERT_EDGE); + if ( $this->edges[VERT_EDGE][$row][$col+1] ) $neigh[$n++] = array($row, $col+1,VERT_EDGE); + + if ( $n == 2 ) { + $n1=0; $n2=1; + $this->isobarCoord[$isobar][$ncoord++] = array( + $this->getCrossingCoord($neigh[$n1][0],$neigh[$n1][1],$neigh[$n1][2],$ib), + $this->getCrossingCoord($neigh[$n2][0],$neigh[$n2][1],$neigh[$n2][2],$ib) ); + } + elseif ( $n == 4 ) { + // We must determine how to connect the edges either northwest->southeast or + // northeast->southwest. We do that by calculating the imaginary middle value of + // the cell by averaging the for corners. This will compared with the value of the + // top left corner will help determine the orientation of the ridge/creek + $midval = ($this->dataPoints[$row][$col]+$this->dataPoints[$row][$col+1]+$this->dataPoints[$row+1][$col]+$this->dataPoints[$row+1][$col+1])/4; + $v = $this->dataPoints[$row][$col]; + if( $midval == $ib ) { + // Orientation "+" + $n1=0; $n2=1; $n3=2; $n4=3; + } elseif ( ($midval > $ib && $v > $ib) || ($midval < $ib && $v < $ib) ) { + // Orientation of ridge/valley = "\" + $n1=0; $n2=3; $n3=2; $n4=1; + } elseif ( ($midval > $ib && $v < $ib) || ($midval < $ib && $v > $ib) ) { + // Orientation of ridge/valley = "/" + $n1=0; $n2=2; $n3=3; $n4=1; + } + + $this->isobarCoord[$isobar][$ncoord++] = array( + $this->getCrossingCoord($neigh[$n1][0],$neigh[$n1][1],$neigh[$n1][2],$ib), + $this->getCrossingCoord($neigh[$n2][0],$neigh[$n2][1],$neigh[$n2][2],$ib) ); + + $this->isobarCoord[$isobar][$ncoord++] = array( + $this->getCrossingCoord($neigh[$n3][0],$neigh[$n3][1],$neigh[$n3][2],$ib), + $this->getCrossingCoord($neigh[$n4][0],$neigh[$n4][1],$neigh[$n4][2],$ib) ); + + } + } + } + } + + if( count($this->isobarColors) == 0 ) { + // No manually specified colors. Calculate them automatically. + $this->CalculateColors(); + } + return array( $this->isobarCoord, $this->isobarValues, $this->isobarColors ); + } +} + + +/** + * This class represent a plotting of a contour outline of data given as a X-Y matrice + * + */ +class ContourPlot extends Plot { + + private $contour, $contourCoord, $contourVal, $contourColor; + private $nbrCountours = 0 ; + private $dataMatrix = array(); + private $invertLegend = false; + private $interpFactor = 1; + private $flipData = false; + private $isobar = 10; + private $showLegend = false; + private $highcontrast = false, $highcontrastbw = false; + private $manualIsobarColors = array(); + + /** + * Construct a contour plotting algorithm. The end result of the algorithm is a sequence of + * line segments for each isobar given as two vertices. + * + * @param $aDataMatrix The Z-data to be used + * @param $aIsobar A mixed variable, if it is an integer then this specified the number of isobars to use. + * The values of the isobars are automatically detrmined to be equ-spaced between the min/max value of the + * data. If it is an array then it explicetely gives the isobar values + * @param $aInvert By default the matrice with row index 0 corresponds to Y-value 0, i.e. in the bottom of + * the plot. If this argument is true then the row with the highest index in the matrice corresponds to + * Y-value 0. In affect flipping the matrice around an imaginary horizontal axis. + * @param $aHighContrast Use high contrast colors (blue/red:ish) + * @param $aHighContrastBW Use only black colors for contours + * @return an instance of the contour plot algorithm + */ + function __construct($aDataMatrix, $aIsobar=10, $aFactor=1, $aInvert=false, $aIsobarColors=array()) { + + $this->dataMatrix = $aDataMatrix; + $this->flipData = $aInvert; + $this->isobar = $aIsobar; + $this->interpFactor = $aFactor; + + if ( $this->interpFactor > 1 ) { + + if( $this->interpFactor > 5 ) { + JpGraphError::RaiseL(28007);// ContourPlot interpolation factor is too large (>5) + } + + $ip = new MeshInterpolate(); + $this->dataMatrix = $ip->Linear($this->dataMatrix, $this->interpFactor); + } + + $this->contour = new Contour($this->dataMatrix,$this->isobar,$aIsobarColors); + + if( is_array($aIsobar) ) + $this->nbrContours = count($aIsobar); + else + $this->nbrContours = $aIsobar; + } + + + /** + * Flipe the data around the center + * + * @param $aFlg + * + */ + function SetInvert($aFlg=true) { + $this->flipData = $aFlg; + } + + /** + * Set the colors for the isobar lines + * + * @param $aColorArray + * + */ + function SetIsobarColors($aColorArray) { + $this->manualIsobarColors = $aColorArray; + } + + /** + * Show the legend + * + * @param $aFlg true if the legend should be shown + * + */ + function ShowLegend($aFlg=true) { + $this->showLegend = $aFlg; + } + + + /** + * @param $aFlg true if the legend should start with the lowest isobar on top + * @return unknown_type + */ + function Invertlegend($aFlg=true) { + $this->invertLegend = $aFlg; + } + + /* Internal method. Give the min value to be used for the scaling + * + */ + function Min() { + return array(0,0); + } + + /* Internal method. Give the max value to be used for the scaling + * + */ + function Max() { + return array(count($this->dataMatrix[0])-1,count($this->dataMatrix)-1); + } + + /** + * Internal ramewrok method to setup the legend to be used for this plot. + * @param $aGraph The parent graph class + */ + function Legend($aGraph) { + + if( ! $this->showLegend ) + return; + + if( $this->invertLegend ) { + for ($i = 0; $i < $this->nbrContours; $i++) { + $aGraph->legend->Add(sprintf('%.1f',$this->contourVal[$i]), $this->contourColor[$i]); + } + } + else { + for ($i = $this->nbrContours-1; $i >= 0 ; $i--) { + $aGraph->legend->Add(sprintf('%.1f',$this->contourVal[$i]), $this->contourColor[$i]); + } + } + } + + + /** + * Framework function which gets called before the Stroke() method is called + * + * @see Plot#PreScaleSetup($aGraph) + * + */ + function PreScaleSetup($aGraph) { + $xn = count($this->dataMatrix[0])-1; + $yn = count($this->dataMatrix)-1; + + $aGraph->xaxis->scale->Update($aGraph->img,0,$xn); + $aGraph->yaxis->scale->Update($aGraph->img,0,$yn); + + $this->contour->SetInvert($this->flipData); + list($this->contourCoord,$this->contourVal,$this->contourColor) = $this->contour->getIsobars(); + } + + /** + * Use high contrast color schema + * + * @param $aFlg True, to use high contrast color + * @param $aBW True, Use only black and white color schema + */ + function UseHighContrastColor($aFlg=true,$aBW=false) { + $this->highcontrast = $aFlg; + $this->highcontrastbw = $aBW; + $this->contour->UseHighContrastColor($this->highcontrast,$this->highcontrastbw); + } + + /** + * Internal method. Stroke the contour plot to the graph + * + * @param $img Image handler + * @param $xscale Instance of the xscale to use + * @param $yscale Instance of the yscale to use + */ + function Stroke($img,$xscale,$yscale) { + + if( count($this->manualIsobarColors) > 0 ) { + $this->contourColor = $this->manualIsobarColors; + if( count($this->manualIsobarColors) != $this->nbrContours ) { + JpGraphError::RaiseL(28002); + } + } + + $img->SetLineWeight($this->line_weight); + + for ($c = 0; $c < $this->nbrContours; $c++) { + + $img->SetColor( $this->contourColor[$c] ); + + $n = count($this->contourCoord[$c]); + $i = 0; + while ( $i < $n ) { + list($x1,$y1) = $this->contourCoord[$c][$i][0]; + $x1t = $xscale->Translate($x1); + $y1t = $yscale->Translate($y1); + + list($x2,$y2) = $this->contourCoord[$c][$i++][1]; + $x2t = $xscale->Translate($x2); + $y2t = $yscale->Translate($y2); + + $img->Line($x1t,$y1t,$x2t,$y2t); + } + + } + } + +} + +// EOF +?> |