<DIV>Dear,<BR>     I am very interested in your research work and honored to discuss the ASIFT algorithm with you. We can clearly believe that the ASIFT is superior than the SIFT through the online demo. I test my own data using this method and it works too. But there is a details we should discuss. I have downloaded your source code(compilable on Windows systems) to test my data and found the number of key points calculated is different from the online demo for the same images. The points matched are different too. </DIV> <DIV>     So I guess whether the source code in the web-page is the same with the online demo, or you have fogotten update the source code?<BR>     Looking forward to your reply.</DIV> <DIV>     For example,the results of the images in accessories compared by two ways as follows:</DIV> <DIV><FONT style="BACKGROUND-COLOR: #ffffff">     online demo:</FONT></DIV> <BLOCKQUOTE style="MARGIN-RIGHT: 0px" dir="ltr"> <DIV><FONT style="BACKGROUND-COLOR: #c0c0c0">Computing keypoints on the two images...<BR>5076 ASIFT keypoints are detected. <BR>4199 ASIFT keypoints are detected. <BR>Keypoints computation accomplished in 4 seconds.<BR>Matching the keypoints...<BR>The two images match! 75 matchings are identified. log(nfa)=-22.2367.<BR>Keypoints matching accomplished in 6 seconds.</FONT></DIV></BLOCKQUOTE> <DIV dir="ltr">    download source code:</DIV> <BLOCKQUOTE style="MARGIN-RIGHT: 0px" dir="ltr"> <DIV dir="ltr"><FONT style="BACKGROUND-COLOR: #c0c0c0">Computing keypoints on the two images... <BR>49 ASIFT keypoints are detected.  <BR>71 ASIFT keypoints are detected.  <BR>Keypoints computation accomplished in 1 seconds. <BR>Matching the keypoints... <BR>The two images do not match. Not enough matches to do epipolar filtering. <BR>Keypoints matching accomplished in 0 seconds.</FONT></DIV></BLOCKQUOTE><br><br><span title="neteasefooter"><span id="netease_mail_footer"></span></span>