From: Pavel Charvat <pavel.charvat@netcentrum.cz>
Date: Mon, 7 Aug 2006 12:13:54 +0000 (+0200)
Subject: scanner generates tree new files - image signatures (for similarity detection),
X-Git-Tag: holmes-import~611
X-Git-Url: http://mj.ucw.cz/gitweb/?a=commitdiff_plain;h=32a626b1e7d24c528b5c863d79713c3f3f62a068;p=libucw.git

scanner generates tree new files - image signatures (for similarity detection),
small and big thumbnails (for duplicates search)
---

diff --git a/images/object.c b/images/object.c
index 0d6a146f..3935f89b 100644
--- a/images/object.c
+++ b/images/object.c
@@ -52,7 +52,7 @@ get_image_obj_thumb(struct image_obj_info *ioi, struct odes *o, struct mempool *
     count++;
   byte buf[count * MAX_ATTR_SIZE], *b = buf;
   for (; a; a = a->same)
-    b += base224_decode(buf, a->val, strlen(a->val));
+    b += base224_decode(b, a->val, strlen(a->val));
   ASSERT(b != buf);
   ioi->thumb_data = mp_alloc(pool, ioi->thumb_size = b - buf);
   memcpy(ioi->thumb_data, buf, ioi->thumb_size);
@@ -89,20 +89,20 @@ void
 put_image_obj_signature(struct odes *o, struct image_signature *sig)
 {
   /* signatures should be short enough to fit one attribute */
-  ASSERT(MAX_ATTR_SIZE > BASE224_ENC_LENGTH(sizeof(struct image_vector) + 4 + sig->len * sizeof(struct image_region)));
-  byte buf[MAX_ATTR_SIZE], *b = buf;
-  memcpy(b, &sig->vec, sizeof(struct image_vector));
-  b += sizeof(struct image_vector);
-  *b++ = sig->len;
-  *b++ = sig->df;
-  *(u16 *)b++ = sig->dh;
-  for (uns i = 0; i < sig->len; i++)
-    {
-      memcpy(b, sig->reg + i, sizeof(struct image_region));
-      b += sizeof(struct image_region);
-    }
-  uns len = b - buf;
-  byte b224[MAX_ATTR_SIZE];
-  b224[base224_encode(b224, buf, len)] = 0;
-  obj_set_attr(o, 'H', b224);
+  byte buf[MAX_ATTR_SIZE];
+  uns size = image_signature_size(sig->len);
+  ASSERT(MAX_ATTR_SIZE > BASE224_ENC_LENGTH(size));
+  buf[base224_encode(buf, (byte *)sig, size)] = 0;
+  obj_set_attr(o, 'H', buf);
+}
+
+uns
+get_image_obj_signature(struct image_signature *sig, struct odes *o)
+{
+  byte *a = obj_find_aval(o, 'H');
+  if (!a)
+    return 0;
+  UNUSED uns size = base224_decode((byte *)sig, a, strlen(a));
+  ASSERT(size == image_signature_size(sig->len));
+  return 1;
 }
diff --git a/images/object.h b/images/object.h
index 58e789d6..189b3c0b 100644
--- a/images/object.h
+++ b/images/object.h
@@ -22,5 +22,6 @@ uns get_image_obj_info(struct image_obj_info *ioi, struct odes *o);
 uns get_image_obj_thumb(struct image_obj_info *ioi, struct odes *o, struct mempool *pool);
 struct image *read_image_obj_thumb(struct image_obj_info *ioi, struct fastbuf *fb, struct image_io *io, struct mempool *pool);
 void put_image_obj_signature(struct odes *o, struct image_signature *sig);
+uns get_image_obj_signature(struct image_signature *sig, struct odes *o);
 
 #endif
diff --git a/images/sig-init.c b/images/sig-init.c
index 17250e68..091d7d6c 100644
--- a/images/sig-init.c
+++ b/images/sig-init.c
@@ -19,22 +19,6 @@
 
 static double image_sig_inertia_scale[3] = { 3, 1, 0.3 };
 
-void
-bget_image_signature(struct fastbuf *fb, struct image_signature *sig)
-{
-  breadb(fb, &sig->vec, sizeof(sig->vec));
-  sig->len = bgetc(fb);
-  breadb(fb, sig->reg, sig->len * sizeof(*sig->reg));
-}
-
-void
-bput_image_signature(struct fastbuf *fb, struct image_signature *sig)
-{
-  bwrite(fb, &sig->vec, sizeof(sig->vec));
-  bputc(fb, sig->len);
-  bwrite(fb, sig->reg, sig->len * sizeof(*sig->reg));
-}
-
 struct block {
   u32 l, u, v;		/* average Luv coefficients */
   u32 lh, hl, hh;	/* energies in Daubechies wavelet bands */
diff --git a/images/signature.h b/images/signature.h
index 2c2ab101..695ca722 100644
--- a/images/signature.h
+++ b/images/signature.h
@@ -6,27 +6,33 @@
 #define IMAGE_REG_H	3
 #define IMAGE_REG_MAX	8
 
-/* K-dimensional feature vector */
+/* K-dimensional feature vector (6 bytes) */
 struct image_vector {
   byte f[IMAGE_VEC_F];		/* texture features */
 } PACKED;
 
-/* Fetures for image regions */
+/* Fetures for image regions (16 bytes) */
 struct image_region {
   byte f[IMAGE_VEC_F];		/* texture features */
   u16 h[IMAGE_REG_H];		/* shape features */
-  byte wa;			/* normalized area percentage */
-  byte wb;			/* normalized weight */
+  u16 wa;			/* normalized area percentage */
+  u16 wb;			/* normalized weight */
 } PACKED;
 
-/* Image signature */
+/* Image signature (10 + len * 16 bytes) */
 struct image_signature {
-  struct image_vector vec;	/* Combination of all regions... simple signature */
   byte len;			/* Number of regions */
   byte df;			/* average f dist */
   u16 dh;			/* average h dist */
+  struct image_vector vec;	/* Combination of all regions... simple signature */
   struct image_region reg[IMAGE_REG_MAX];/* Feature vector for every region */
-};
+} PACKED;
+
+static inline uns
+image_signature_size(uns len)
+{
+  return 4 + sizeof(struct image_vector) + len * sizeof(struct image_region);
+}
 
 /* sig-dump.c */
 
@@ -38,8 +44,6 @@ byte *image_region_dump(byte *buf, struct image_region *reg);
 
 /* sig-init.c */
 
-void bget_image_signature(struct fastbuf *fb, struct image_signature *sig);
-void bput_image_signature(struct fastbuf *fb, struct image_signature *sig);
 int compute_image_signature(struct image_thread *thread, struct image_signature *sig, struct image *image);
 
 /* sig-cmp.c */