Conclusion

Final Report/report.tex

@@ -331,6 +331,9 @@ \subsection{Discussion}
 Finally, we also note that adjusting the weights is critically important for a good result. This issue is a keypoint for any Future Work on this project.
 
 \section{Conclusion}
+When we look at our results of the OCR application, we see that some improvements are needed if we want the program to work perfectly. The error rate should be reduced significantly, and this can be tried by different approaches. We can do this by increasing the test phase, using more variations as input for our program. Different, and more decisive features can be implemented, further decreasing the error rate. It is questionable though that very similar characters like 0 and D can ever be 100\% correctly recognized. Finally, we could optimize the digitization process of text input even further by implementing a dictonary. This could be seen as a 'second opinion' on the decision made by our application. If OCR is rather doubtfull about the character, we can lookup the possible combinations in the dictionary and pick the most plausible one.
+
+However, we can conclude that OCR is a very possible application (as has been proved by commercial applications): letters and numbers can be recognized based on their characteristics and in very reasonable time. This observation has lead to a series of applications found nowadays in our daily life. With the increase of computation power and portability of our devices, we can only predict these applications to become even more popular.
 
 \section{Division of Labor}