.\" .\" Copyright (c) 2007-2014 Frank DENIS .\" .\" Permission to use, copy, modify, and distribute this software for any .\" purpose with or without fee is hereby granted, provided that the above .\" copyright notice and this permission notice appear in all copies. .\" .\" THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES .\" WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF .\" MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR .\" ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES .\" WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN .\" ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF .\" OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. .\" .Dd $Mdocdate: March 31 2011 $ .Dt LIBPUZZLE 3 .Sh NAME .Nm puzzle_init_cvec , .Nm puzzle_init_dvec , .Nm puzzle_fill_dvec_from_file , .Nm puzzle_fill_cvec_from_file , .Nm puzzle_fill_dvec_from_mem , .Nm puzzle_fill_cvec_from_mem , .Nm puzzle_fill_cvec_from_dvec , .Nm puzzle_free_cvec , .Nm puzzle_free_dvec , .Nm puzzle_init_compressed_cvec , .Nm puzzle_free_compressed_cvec , .Nm puzzle_compress_cvec , .Nm puzzle_uncompress_cvec , .Nm puzzle_vector_normalized_distance .Nd compute comparable signatures of bitmap images. .Sh SYNOPSIS .Fd #include .Ft void .Fn puzzle_init_context "PuzzleContext *context" .Ft void .Fn puzzle_free_context "PuzzleContext *context" .Ft void .Fn puzzle_init_cvec "PuzzleContext *context" "PuzzleCvec *cvec" .Ft void .Fn puzzle_init_dvec "PuzzleContext *context" "PuzzleDvec *dvec" .Ft int .Fn puzzle_fill_dvec_from_file "PuzzleContext *context" "PuzzleDvec * dvec" "const char *file" .Ft int .Fn puzzle_fill_cvec_from_file "PuzzleContext *context" "PuzzleCvec * cvec" "const char *file" .Ft int .Fn puzzle_fill_dvec_from_mem "PuzzleContext *context" "PuzzleDvec * dvec" "const void *mem" "size_t size" .Ft int .Fn puzzle_fill_cvec_from_mem "PuzzleContext *context" "PuzzleCvec * cvec" "const void *mem" "size_t size" .Ft int .Fn puzzle_fill_cvec_from_dvec "PuzzleContext *context" "PuzzleCvec * cvec" "const PuzzleDvec *dvec" .Ft void .Fn puzzle_free_cvec "PuzzleContext *context" "PuzzleCvec *cvec" .Ft void .Fn puzzle_free_dvec "PuzzleContext *context" "PuzzleDvec *dvec" .Ft void .Fn puzzle_init_compressed_cvec "PuzzleContext *context" "PuzzleCompressedCvec * compressed_cvec" .Ft void .Fn puzzle_free_compressed_cvec "PuzzleContext *context" "PuzzleCompressedCvec * compressed_cvec" .Ft int .Fn puzzle_compress_cvec "PuzzleContext *context" "PuzzleCompressedCvec * compressed_cvec" "const PuzzleCvec * cvec" .Ft int .Fn puzzle_uncompress_cvec "PuzzleContext *context" "PuzzleCompressedCvec * compressed_cvec" "PuzzleCvec * const cvec" .Ft double .Fn puzzle_vector_normalized_distance "PuzzleContext *context" "const PuzzleCvec * cvec1" "const PuzzleCvec * cvec2" "int fix_for_texts" .Sh DESCRIPTION The Puzzle library computes a signature out of a bitmap picture. Signatures are comparable and similar pictures have similar signatures. .Pp After a picture has been loaded and uncompressed, featureless parts of the image are skipped (autocrop), unless that step has been explicitely disabled, see .Xr puzzle_set 3 .Sh LIBPUZZLE CONTEXT Every public function requires a .Va PuzzleContext object, that stores every required tunables. .Pp Any application using libpuzzle should initialize a .Va PuzzleContext object with .Fn puzzle_init_context and free it after use with .Fn puzzle_free_context .Bd \-literal \-offset indent PuzzleContext context; puzzle_init_context(&context); ... puzzle_free_context(&context); .Ed .Sh DVEC AND CVEC VECTORS The next step is to divide the cropped image into a grid and to compute the average intensity of soft\(hyedged pixels in every block. The result is a .Va PuzzleDvec object. .Pp .Va PuzzleDvec objects should be initialized before use, with .Fn puzzle_init_dvec and freed after use with .Fn puzzle_free_dvec .Pp The .Va PuzzleDvec structure has two important fields: .Va vec is the pointer to the first element of the array containing the average intensities, and .Va sizeof_compressed_vec is the number of elements. .Pp .Va PuzzleDvec objects are not comparable, so what you usually want is to transform these objects into .Va PuzzleCvec objects. .Pp A .Va PuzzleCvec object is a vector with relationships between adjacent blocks from a .Va PuzzleDvec object. .Pp The .Fn puzzle_fill_cvec_from_dvec fills a .Va PuzzleCvec object from a .Va PuzzleDvec object. .Pp But just like the other structure, .Va PuzzleCvec objects must be initialized and freed with .Fn puzzle_init_cvec and .Fn puzzle_free_cvec .Pp .Va PuzzleCvec objects have a vector whoose first element is in the .Va vec field, and the number of elements is in the .Va sizeof_vec field .Sh LOADING PICTURES .Va PuzzleDvec and .Va PuzzleCvec objects can be computed from a bitmap picture file, with .Fn puzzle_fill_dvec_from_file and .Fn puzzle_fill_cvec_from_file .Pp .Em GIF , .Em PNG and .Em JPEG files formats are currently supported and automatically recognized. .Pp Here's a simple example that creates a .Va PuzzleCvec objects out of a file. .Bd \-literal \-offset indent PuzzleContext context; PuzzleCvec cvec; puzzle_init_context(&context); puzzle_init_cvec(&context, &cvec); puzzle_fill_cvec_from_file(&context, &cvec, "test\-picture.jpg"); ... puzzle_free_cvec(&context, &cvec); puzzle_free_context(&context); .Ed .Sh COMPARING VECTORS In order to check whether two pictures are similar, you need to compare their .Va PuzzleCvec signatures, using .Fn puzzle_vector_normalized_distance .Pp That function returns a distance, between 0.0 and 1.0. The lesser, the nearer. .Pp Tests on common pictures show that a normalized distance of 0.6 (also defined as .Va PUZZLE_CVEC_SIMILARITY_THRESHOLD ) means that both pictures are visually similar. .Pp If that threshold is not right for your set of pictures, you can experiment with .Va PUZZLE_CVEC_SIMILARITY_HIGH_THRESHOLD , .Va PUZZLE_CVEC_SIMILARITY_LOW_THRESHOLD and .Va PUZZLE_CVEC_SIMILARITY_LOWER_THRESHOLD or with your own value. .Pp If the .Fa fix_for_texts of .Fn puzzle_vector_normalized_distance is .Em 1 , a fix is applied to the computation in order to deal with bitmap pictures that contain text. That fix is recommended, as it allows using the same threshold for that kind of picture as for generic pictures. .Pp If .Fa fix_for_texts is .Em 0 , that special way of computing the normalized distance is disabled. .Bd \-literal \-offset indent PuzzleContext context; PuzzleCvec cvec1, cvec2; double d; puzzle_init_context(&context); puzzle_init_cvec(&context, &cvec1); puzzle_init_cvec(&context, &cvec2); puzzle_fill_cvec_from_file(&context, &cvec1, "test\-picture\-1.jpg"); puzzle_fill_cvec_from_file(&context, &cvec2, "test\-picture\-2.jpg"); d = puzzle_vector_normalized_distance(&context, &cvec1, &cvec2, 1); if (d < PUZZLE_CVEC_SIMILARITY_THRESHOLD) { puts("Pictures are similar"); } puzzle_free_cvec(&context, &cvec2); puzzle_free_cvec(&context, &cvec1); puzzle_free_context(&context); .Ed .Sh CVEC COMPRESSION In order to reduce storage needs, .Va PuzzleCvec objects can be compressed to 1/3 of their original size. .Pp .Va PuzzleCompressedCvec structures hold the compressed data. Before and after use, these structures have to be passed to .Fn puzzle_init_compressed_cvec and .Fn puzzle_free_compressed_cvec .Pp .Fn puzzle_compress_cvec compresses a .Va PuzzleCvec object into a .Va PuzzleCompressedCvec object. .Pp And .Fn puzzle_uncompress_cvec uncompresses a .Va PuzzleCompressedCvec object into a .Va PuzzleCvec object. .Bd \-literal \-offset indent PuzzleContext context; PuzzleCvec cvec; PuzzleCompressedCvec c_cvec; ... puzzle_init_compressed_cvec(&context, &c_cvec); puzzle_compress_cvec(&context, &c_cvec, &cvec); ... puzzle_free_compressed_cvec(&context, &c_cvec); .Ed The .Va PuzzleCompressedCvec structure has two important fields: .Va vec that is a pointer to the first element of the compressed data, and .Va sizeof_compressed_vec that contains the number of elements. .Sh RETURN VALUE Functions return .Em 0 on success, and .Em \-1 if something went wrong. .Sh AUTHORS .Nf Frank DENIS libpuzzle at pureftpd dot org .Fi .Sh ACKNOWLEDGMENTS .Nf Xerox Research Center H. CHI WONG Marschall BERN David GOLDBERG Sameh SCHAFIK .Fi .Sh SEE ALSO .Xr puzzle_set 3 .Xr puzzle\-diff 8