/* * This file is part of Aptdec. * Copyright (c) 2004-2009 Thierry Leconte (F4DWV), Xerbo (xerbo@protonmail.com) 2019 * * Aptdec is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by * the Free Software Foundation, either version 2 of the License, or * (at your option) any later version. * * This program is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the * GNU General Public License for more details. * * You should have received a copy of the GNU General Public License * along with this program. If not, see . * */ #include #include #include #include #include "offsets.h" #include "messages.h" #define REGORDER 3 typedef struct { double cf[REGORDER + 1]; } rgparam; extern void polyreg(const int m, const int n, const double x[], const double y[], double c[]); // Compute regression static void rgcomp(double x[16], rgparam * rgpr) { // { 0.106, 0.215, 0.324, 0.433, 0.542, 0.652, 0.78, 0.87, 0.0 } const double y[9] = { 31.07, 63.02, 94.96, 126.9, 158.86, 191.1, 228.62, 255.0, 0.0 }; polyreg(REGORDER, 9, x, y, rgpr -> cf); } // Convert a value to 0-255 based off the provided regression curve static double rgcal(float x, rgparam *rgpr) { double y, p; int i; for (i = 0, y = 0.0, p = 1.0; i < REGORDER + 1; i++) { y += rgpr->cf[i] * p; p = p * x; } return(y); } static double tele[16]; static double Cs; static int nbtele; // Brightness equalise, including telemetry void equalise(float **prow, int nrow, int offset, int width, int telestart, rgparam regr[30]){ offset -= SYNC_WIDTH+SPC_WIDTH; for (int n = 0; n < nrow; n++) { float *pixelv = prow[n]; for (int i = 0; i < width+SYNC_WIDTH+SPC_WIDTH+TELE_WIDTH; i++) { float pv = pixelv[i + offset]; // Blend between the calculated regression curves /* TODO: this can actually make the image look *worse* * if the signal has a constant input gain. */ int k, kof; k = (n - telestart) / FRAME_LEN; if (k >= nbtele) k = nbtele - 1; kof = (n - telestart) % FRAME_LEN; if (kof < 64) { if (k < 1) { pv = rgcal(pv, &(regr[k])); } else { pv = rgcal(pv, &(regr[k])) * (64 + kof) / FRAME_LEN + rgcal(pv, &(regr[k - 1])) * (64 - kof) / FRAME_LEN; } } else { if ((k + 1) >= nbtele) { pv = rgcal(pv, &(regr[k])); } else { pv = rgcal(pv, &(regr[k])) * (192 - kof) / FRAME_LEN + rgcal(pv, &(regr[k + 1])) * (kof - 64) / FRAME_LEN; } } pv = CLIP(pv, 0, 255); pixelv[i + offset] = pv; } } } // Get telemetry data for thermal calibration/equalization int calibrate(float **prow, int nrow, int offset, int width, int contrastEqualise) { double teleline[3000]; double wedge[16]; rgparam regr[30]; int n, k; int mtelestart = 0, telestart; int channel = -1; // Calculate average of a row of telemetry for (n = 0; n < nrow; n++) { float *pixelv = prow[n]; // Average the center 40px teleline[n] = 0.0; for (int i = 3; i < 43; i++) teleline[n] += pixelv[i + offset + width]; teleline[n] /= 40.0; } // The minimum rows required to decode a full frame if (nrow < 192) { fprintf(stderr, ERR_TELE_ROW); return(0); } /* Wedge 7 is white and 8 is black, which will have the largest * difference in brightness, this will always be in the center of * the frame and can thus be used to find the start of the frame */ float max = 0; for (n = nrow / 3 - 64; n < 2 * nrow / 3 - 64; n++) { float df; // (sum 4px below) / (sum 4px above) df = (teleline[n - 4] + teleline[n - 3] + teleline[n - 2] + teleline[n - 1]) / (teleline[n + 0] + teleline[n + 1] + teleline[n + 2] + teleline[n + 3]); // Find the maximum difference if (df > max) { mtelestart = n; max = df; } } // Find the start of the first frame telestart = (mtelestart - FRAME_LEN/2) % FRAME_LEN; // Make sure that theres at least one full frame in the image if (nrow < telestart + FRAME_LEN) { fprintf(stderr, ERR_TELE_ROW); return(0); } // For each frame for (n = telestart, k = 0; n < nrow - FRAME_LEN; n += FRAME_LEN, k++) { float *pixelv = prow[n]; int j; // Turn each wedge into a value for (j = 0; j < 16; j++) { // Average the middle 6px wedge[j] = 0.0; for (int i = 1; i < 7; i++) wedge[j] += teleline[(j * 8) + n + i]; wedge[j] /= 6; } // Compute regression on the wedges rgcomp(wedge, &(regr[k])); // Read the telemetry values from the middle of the image if (k == nrow / (2*FRAME_LEN)) { int l; // Equalise for (j = 0; j < 16; j++) tele[j] = rgcal(wedge[j], &(regr[k])); /* Compare the channel ID wedge to the reference * wedges, the wedge with the closest match will * be the channel ID */ float min = 10000; for (j = 0; j < 6; j++) { float df; df = tele[15] - tele[j]; df *= df; if (df < min) { channel = j; min = df; } } // Cs computation, still have no idea what this does int i; for (Cs = 0.0, i = 0, j = n; j < n + FRAME_LEN; j++) { double csline; for (csline = 0.0, l = 3; l < 43; l++) csline += pixelv[l + offset - SPC_WIDTH]; csline /= 40.0; if (csline > 50.0) { Cs += csline; i++; } } Cs /= i; Cs = rgcal(Cs, &(regr[k])); } } nbtele = k; if(contrastEqualise) equalise(prow, nrow, offset, width, telestart, regr); return(channel + 1); } // --- Temperature Calibration --- // extern int satnum; #include "satcal.h" typedef struct { double Nbb; double Cs; double Cb; int ch; } tempparam; // IR channel temperature compensation static void tempcomp(double t[16], int ch, tempparam *tpr) { double Tbb, T[4]; double C; tpr -> ch = ch - 4; // Compute equivalent T blackbody temperature for (int n = 0; n < 4; n++) { float d0, d1, d2; C = t[9 + n] * 4.0; d0 = satcal[satnum].d[n][0]; d1 = satcal[satnum].d[n][1]; d2 = satcal[satnum].d[n][2]; T[n] = d0; T[n] += d1 * C; C = C * C; T[n] += d2 * C; } Tbb = (T[0] + T[1] + T[2] + T[3]) / 4.0; Tbb = satcal[satnum].rad[tpr->ch].A + satcal[satnum].rad[tpr->ch].B * Tbb; // Compute radiance blackbody C = satcal[satnum].rad[tpr->ch].vc; tpr->Nbb = c1 * C * C * C / (expm1(c2 * C / Tbb)); // Store count blackbody and space tpr->Cs = Cs * 4.0; tpr->Cb = t[14] * 4.0; } // IR channel temperature calibration static double tempcal(float Ce, tempparam * rgpr) { double Nl, Nc, Ns, Ne; double T, vc; Ns = satcal[satnum].cor[rgpr->ch].Ns; Nl = Ns + (rgpr->Nbb - Ns) * (rgpr->Cs - Ce * 4.0) / (rgpr->Cs - rgpr->Cb); Nc = satcal[satnum].cor[rgpr->ch].b[0] + satcal[satnum].cor[rgpr->ch].b[1] * Nl + satcal[satnum].cor[rgpr->ch].b[2] * Nl * Nl; Ne = Nl + Nc; vc = satcal[satnum].rad[rgpr->ch].vc; T = c2 * vc / log1p(c1 * vc * vc * vc / Ne); T = (T - satcal[satnum].rad[rgpr->ch].A) / satcal[satnum].rad[rgpr->ch].B; // Rescale to 0-255 for -60'C to +40'C T = (T - 273.15 + 60.0) / 100.0 * 256.0; return(T); } // Temperature calibration wrapper void temperature(float **prow, int nrow, int channel, int offset) { tempparam temp; printf("Temperature... "); fflush(stdout); tempcomp(tele, channel, &temp); for (int n = 0; n < nrow; n++) { float *pixelv = prow[n]; for (int i = 0; i < CH_WIDTH; i++) { float pv = tempcal(pixelv[i + offset], &temp); pv = CLIP(pv, 0, 255); pixelv[i + offset] = pv; } } printf("Done\n"); }