image similarity metrics

image_list.py

@@ -30,10 +30,8 @@ def setup_options():
                      help="Create randomized files for landscape and portrait images [%(default)s]")
     parser.add_argument("-s",type=str,dest="search",default=False,
                       help="Search list based on path pattern")
-    parser.add_argument("--color",action="store_true",dest="colors",default=False,
+    parser.add_argument("--measure",action="store_true",dest="measure",default=False,
-                     help="Append list with mean color information This option will flip the 'Add new files' option. [%(default)s]")
+                     help="Measure various statistics for similarity/color searches. This option will flip the 'Add new files' option. [%(default)s]")
-    parser.add_argument("--finger",action="store_true",dest="finger",default=False,
-                     help="Append list with fingerprint information. This option will flip the 'Add new files' option. [%(default)s]")
     parser.add_argument("--nearest",type=str,dest="nearestcolor",default=False,
                      help="Search list for nearest mean color. format: R,G,B in float 0-1. Add fourth value to limit search to number")
     parser.add_argument("--dup",action="store_true",dest="duplicate",default=False,
@@ -42,7 +40,7 @@ def setup_options():
                      help="Delete files listed with --small. [%(default)s]")
     parser.add_argument("--small",action="store_true",dest="searchsmall",default=False,
                      help="Return a list of small files, smaller than -m INT. This option will flip the 'Add new files' option. [%(default)s]")
-    parser.add_argument("--similar",type=int,dest="similarity",default=False,
+    parser.add_argument("--similar",type=str,dest="similarity",default=None,
                      help="Search list for similar images. Value 0-255 for similarity threshold. 0=high similarity")
     parser.add_argument("-x",action="append",dest="exclude",default=[],
                      help="Exclude folder name from the lists. This option may be issued several times")
@@ -50,7 +48,7 @@ def setup_options():
     
     options=parser.parse_args()
     BADDIRS.extend(options.exclude)
-    if options.duplicate or options.searchsmall or options.colors or options.finger:
+    if options.duplicate or options.searchsmall or options.measure or options.nearestcolor or options.similarity!=None or options.search:
         options.add=not options.add
     return options
 
@@ -298,8 +296,8 @@ def get_fingerprint(filename):
             values+='0'
     return str(int(values,2))
 
-def find_fingerprint_nearest(sqlfile,thr):
+def find_fingerprint_similar(sqlfile,thr):
-    #   thr either file, or threshold?
+    ''' Find all similar images, nearest match more similar than thr '''
     conn=sqlite3.connect(sqlfile)
     conn.text_factory=str
     db1=conn.cursor()
@@ -325,12 +323,33 @@ def find_fingerprint_nearest(sqlfile,thr):
                         this[0]=foo
                     hits.append(hit2[0])
         if this[2]<thr:
-            print('"%s" "%s"' % ( this[0], this[1] ))
+            print('"%s" "%s":%i' % ( this[0], this[1], this[2] ))
-    #for row in hits:
-    #    print(row)
-            
     return
 
+def find_fingerprint_nearest(sqlfile,cmp):
+    ''' Find nearest match '''
+    cmp=os.path.abspath(cmp)
+    conn=sqlite3.connect(sqlfile)
+    conn.text_factory=str
+    db1=conn.cursor()
+    if is_listed(db1, cmp):
+        db1.execute("SELECT file,fingerprint,width,height FROM list WHERE file=?",(cmp,))
+        for hit1 in db1:
+            fp=int(hit1[1])
+    else:
+        fp=int(get_fingerprint(cmp))
+
+    db1.execute("SELECT file,fingerprint,width,height FROM list ORDER BY file")
+    this=['',sys.maxint]
+    for i,hit1 in enumerate(db1):
+        if hit1[0] == cmp:
+            continue
+        similarity=bin(fp^int(hit1[1])).count('1')
+        if similarity<this[1]:
+            this[1]=similarity
+            this[0]=hit1[0]
+    print('"%s":%i' % (this[0], this[1]))
+
 def searchdb(sqlfile,needle):
     conn=sqlite3.connect(sqlfile)
     conn.text_factory=str
@@ -440,30 +459,29 @@ def main():
     options=setup_options();
     if not os.path.exists(options.sqlfile):
         createdb(options.sqlfile);
-    if options.search:
-        print_structure(searchdb(options.sqlfile,options.search))
-        sys.exit(0)
-    if options.nearestcolor:
-        find_color_nearest(options.sqlfile,options.nearestcolor)
-        sys.exit(0)
-    if options.similarity:
-        find_fingerprint_nearest(options.sqlfile,options.similarity)
-        sys.exit(0)
     if options.delete:
         print('Deleting entries...') 
         delete_nonexisting(options.sqlfile)
     if options.add or options.changed:
         print('Adding entries...')
         add_recurse(options)
-    if options.colors:
+    if options.search:
+        print_structure(searchdb(options.sqlfile,options.search))
+    if options.measure:
         print('Adding colors...')
         append_colors(options.sqlfile)
-    if options.finger:
         print('Adding fingerprints...')
         append_fingerprints(options.sqlfile)
     if options.random:
         print('Random lists...')
         random_lists(options.sqlfile)
+    if options.nearestcolor:
+        find_color_nearest(options.sqlfile,options.nearestcolor)
+    if options.similarity!=None:
+        if os.path.exists(options.similarity):
+            find_fingerprint_nearest(options.sqlfile,options.similarity)
+        else: 
+            find_fingerprint_similar(options.sqlfile,int(options.similarity))
     if options.duplicate:
         files=find_duplicates(options.sqlfile,options.startpath)
         print_dup_structure(files)