Invert subtitle image before the OCR

It makes it work way better on my side, especially
with subtitles containing multiple lines.

doc/completion.sh

@@ -44,7 +44,7 @@ _vobsub2srt() {
 
     case $cur in
         -*)
-            COMPREPLY=( $( compgen -W '--dump-images --verbose --ifo --lang --langlist --tesseract-lang --tesseract-data --blacklist --y-threshold --min-width --min-height --max-threads' -- "$cur" ) )
+            COMPREPLY=( $( compgen -W '--dump-images --verbose --ifo --lang --langlist --tesseract-lang --tesseract-data --blacklist --y-threshold --min-width --min-height --max-threads --disable-invert' -- "$cur" ) )
             ;;
         *)
             _filedir '(idx|IDX|sub|SUB)'

doc/vobsub2srt.1

@@ -49,6 +49,8 @@ Minimum height in pixels to consider a subpicture for OCR (Default: 1).
 .TP
 \fB\-\-max\-threads\fR \fInb\fR
 Maximum number of threads to use to do the OCR, use 0 to autodetect the number of cores (Default: 0).
+\fB\-\-disable\-invert\fR
+By default the image will be inverted before the OCR because it works better with black on white background, disable it.
 .SH EXAMPLES
 .nf
   $ \fBvobsub2srt \-\-lang en foobar\fR

src/vobsub2srt.c++

@@ -142,6 +142,16 @@ void do_ocr(TessBaseAPI *tess_base_api, atomic<bool> *done, vector<sub_text_t> *
   done->store(true);
 }
 
+void invert_image(unsigned width, unsigned height,
+                  unsigned stride, unsigned char *image) {
+  for (unsigned y=0; y < height; y++) {
+    for (unsigned x=0; x < width; x++) {
+      unsigned index = y*stride + x;
+      image[index] = 255 - image[index];
+    }
+  }
+}
+
 struct ocr_thread_t {
   ocr_thread_t(TessBaseAPI *tess_base_api)
     : tess_base_api(tess_base_api)
@@ -153,6 +163,7 @@ struct ocr_thread_t {
 
 int main(int argc, char **argv) {
   bool dump_images = false;
+  bool disable_invert = false;
   bool verb = false;
   bool list_languages = false;
   std::string ifo_file;
@@ -186,6 +197,7 @@ int main(int argc, char **argv) {
       add_option("min-width", min_width, "Minimum width in pixels to consider a subpicture for OCR (Default: 9)").
       add_option("min-height", min_height, "Minimum height in pixels to consider a subpicture for OCR (Default: 1)").
       add_option("max-threads", max_threads, "Maximum number of threads to use to do the OCR, use 0 to autodetect the number of cores (Default: 0)").
+      add_option("disable-invert", disable_invert, "By default the image will be inverted before the OCR because it works better with black on white background, disable it").
       add_unnamed(subname, "subname", "name of the subtitle files WITHOUT .idx/.sub ending! (REQUIRED)");
     if(not opts.parse_cmd(argc, argv) or subname.empty()) {
       return 1;
@@ -344,6 +356,9 @@ int main(int argc, char **argv) {
       unsigned char *image_cpy = (unsigned char *)malloc(image_size);
       memcpy(image_cpy, image, image_size);
 
+      if (!disable_invert)
+        invert_image(width, height, stride, image_cpy);
+
       if (max_threads == 1)
         do_ocr(ocr_thread->tess_base_api, &ocr_thread->done, &conv_subs, &mut, sub_counter, width, height, stride, image_cpy, start_pts, end_pts, verb);
       else {