tracker/pt: declare floating-point type size in one place

We want double precision for POSIT. It's best for the type to be set in
ope place without the need to go over everything while switching it back
and forth during tests.

Machine epsilon for float is very small as per
<https://en.wikipedia.org/wiki/Machine_epsilon>. Also see the absurdly
high epsilon of 1e-4 of POSIT that we've had. With floats, making the
epsilon lower resulted in change deltas flushing to zero. This typically
led to the translation Z value being very unstable in PT.

After the epsilon and data type size changes the Z value is stable.

1 files changed, 3 insertions, 3 deletions

diff --git a/tracker-pt/point_extractor.h b/tracker-pt/point_extractor.h
index 94948d34..67b2b8ea 100644
--- a/tracker-pt/point_extractor.h
+++ b/tracker-pt/point_extractor.h
@@ -16,13 +16,13 @@
 
 #include <vector>
 
-class PointExtractor
+class PointExtractor final : private pt_types
 {
 public:
     // extracts points from frame and draws some processing info into frame, if draw_output is set
     // dt: time since last call in seconds
     // WARNING: returned reference is valid as long as object
-    const std::vector<cv::Vec2f> &extract_points(cv::Mat &frame);
+    const std::vector<vec2>& extract_points(cv::Mat &frame);
     int get_n_points() { QMutexLocker l(&mtx); return points.size(); }
     PointExtractor();
 
@@ -31,7 +31,7 @@ private:
     static constexpr double pi = 3.14159265359;
     static constexpr int hist_c = 1;
 
-    std::vector<cv::Vec2f> points;
+    std::vector<vec2> points;
     QMutex mtx;
     cv::Mat frame_gray;
     cv::Mat frame_bin;