first draft of quantize in Likely

Josh Klontz
1 parent 0f880500
Showing 1 changed file with 19 additions and 5 deletions
share/openbr/likely/face_recognition.tex
@@ -16,6 +16,21 @@ This document represents a long-term effort to port the OpenBR face recognition 
 As Likely is a literate programming language, this document is both the source code \emph{and} the documentation.
 \end{abstract}
+\section{Quantization}
+\begin{likely}
+quantize :=
+  training-data :->
+    src :->
+    {
+      dst := (imitate-size src (imitate-dimensions u8 src.type))
+      lo := training-data.min-element
+      hi := training-data.max-element
+      scale := (/ 255 (- hi lo))
+      (dst src lo scale) :=>
+        dst :<- src.(- lo).(* scale)
+    }
+\end{likely}
+
 \section{Consolidated Algorithm}
 The top level definition of the face recognition algorithm.
@@ -23,9 +38,8 @@ The top level definition of the face recognition algorithm.
 face-recognition :=
   src :->
   {
-    dst := src.imitate
-    (dst src) :=>
-      dst :<- src
+    algorithm := (quantize data)
+    (algorithm src)
   }
 \end{likely}
@@ -33,7 +47,7 @@ face-recognition :=
 The entry point to the outside world is a single C function \texttt{face\_recognition} that takes as input a \texttt{likely\_mat} (the input image) and outputs a new \texttt{likely\_mat} (the feature vector).
 \begin{likely}
-(extern u8CXY "face_recognition" u8CXY face-recognition)
+(extern u8X "face_recognition" f32X face-recognition)
 \end{likely}
-\end{document}
 \ No newline at end of file
+\end{document}