06fc0912e4
Actual patch is a mix of work from subitter and my modifications based on debian patch from [1] While here, fix the DCMTK option. PR: 215240 Submitted by: jbeich Obtained From: https://anonscm.debian.org/viewvc/debian-med/trunk/packages/amide/trunk/debian/patches/libav10.patch?view=markup&pathrev=16327 [1]
25 lines
1.2 KiB
C
25 lines
1.2 KiB
C
--- src/alignment_mutual_information.c.orig 2012-01-22 16:25:06 UTC
|
|
+++ src/alignment_mutual_information.c
|
|
@@ -28,6 +28,13 @@
|
|
#include "amitk_data_set_DOUBLE_0D_SCALING.h"
|
|
#include "alignment_mutual_information.h"
|
|
|
|
+#if __FreeBSD__ >= 2
|
|
+#include <osreldate.h>
|
|
+#if __FreeBSD_version <= 802502
|
|
+#define log2(x) (log(x) / M_LN2)
|
|
+#endif
|
|
+#endif
|
|
+
|
|
/* this algorithm will calculate the amount of mutual information between two data sets in their current orientations */
|
|
/* it is a re-write of the original algorithm for purposes of improved speed. the hope is that it won't affect accuracy. */
|
|
/* rather than computing mutual information for the whole volume of data, the algorithm computes it for three orthogonal */
|
|
@@ -209,7 +216,7 @@ gdouble calculate_mutual_information(Ami
|
|
//g_print("\t\%i", mutual_information_array[i][j] ); // for point-wise counts
|
|
// g_print("\t\%4.3f", incremental_mi ); // for point-wise probability
|
|
|
|
- if isinf(incremental_mi) {
|
|
+ if(isinf(incremental_mi)) {
|
|
//count it (because lots and lots of zeroes mean bad registration) and go to the next loop
|
|
mi_nan_count++;
|
|
// this is most often a problem when the incemental MI is infinity, due to divide by zero
|