NEW

what www.tineye.com provides:

• TinEye creates a unique and compact digital signature or 'fingerprint'

• compares this fingerprint to every other image in TinEye index to retrieve matches

• TinEye does not typically find similar images

• it finds exact matches including those that have been cropped, edited or resized

the new approach

• search from web directly

• no predefined image database

• find simillar and exactly same images

• can give statics like extent of matching(eg 10 % matching or 90 % matching)

• more robust

• not a biased search.

how i did it

• get the YQL query for trivial image search

                select * from flickr.photos.search where text="tower"

• get the REST query from YQL

            http://query.yahooapis.com/v1/public/yql?q=select%20*%20from%20flickr.photos.search%20where%20text%3D%22.jpg%22%20%20limit%2010&format=xml

• pars the REST query  to get image url using curl,perl,grep,sed and awk

            curl "http://query.yahooapis.com/v1/public/yql?q=select%20*%20from%20flickr.photos.search%20where%20text%3D%22.jpg%22%20%20%0A&format=xml" 2>/dev/null | perl -pe 's,<,\n<,g' | grep '^<photo'  | sed -e 's/[^"]*"\([^"]*\)"/\1 /g' | awk '{printf("http://farm%d.static.flickr.com/%d/%s_%s.jpg\n", $1, $8, $2, $7);}'

• again pars the table containing URL to wget           

for url in $( curl "http://query.yahooapis.com/v1/public/yql?q=select%20*%20from%20flickr.photos.search%20where%20text%3D%22.jpg%22%20%20%0A&format=xml" 2>/dev/null | perl -pe 's,<,\n<,g' | grep '^<photo'  | sed -e 's/[^"]*"\([^"]*\)"/\1 /g' | awk '{printf("http://farm%d.static.flickr.com/%d/%s_%s.jpg\n", $1, $8, $2, $7);}' ); do

    wget -o file2.jpg $url

done

• put in a bash scrip a.sh

        #!/bin/bash

        echo "...........IMAGE download........"

for url in $( curl "http://query.yahooapis.com/v1/public/yql?q=select%20*%20from%20flickr.photos.search%20where%20text%3D%22.jpg%22%20%20%0A&format=xml" 2>/dev/null | perl -pe 's,<,\n<,g' | grep '^<photo'  | sed -e 's/[^"]*"\([^"]*\)"/\1 /g' | awk '{printf("http://farm%d.static.flickr.com/%d/%s_%s.jpg\n", $1, $8, $2, $7);}' ); do

    wget -o file2.jpg $url

done

• write another bash script to compare images

#!/bin/bash

echo "...........IMAGE SEARCH........"

./compare_image.out # executable file for image comparision

cat head256 joint_histogram.int | display & #shows relative alignment of images

cat head out3.int | display &                         # the two images displayed

cat head out4.int | display &

loop holes

please figure it out and report me

I am already having some of them.