Enhancement: unicode/UTF-8 support with InitFont & PutText

[SteveKrebs]

As a user, I would like PutText to support characters beyond basic ASCII chars to be able to utilize additional characters in writing additional languages onto video frames.

Acceptance would be for either extending the existing initFont / putText APIs or creating new APIs ( if necessary to keep backwards compatibility for the original APIs, example: initFont_UTF/ putText_UTF) that support additional characters.

re: cv::HersheyFonts : There is Hershey font support for additional characters in Greek, Cyrillic, Japanese (kanji, hiragana and katakana) as well as support for mathematics, musical notation, map markers, and meteorological symbols.
re: https://sourceforge.net/projects/hershey9000/files/

[vpisarev]

@SteveKrebs, thanks for raising the issue.

I'm now preparing the patch that replaces Hershey fonts with TrueType/OpenType/Type 1 fonts. The current proposal that some font(s) will be embedded into OpenCV and they provide ASCII + Latin extended + Cyrillic + Greek glyphs. And users can also load their own fonts with arbitrary Unicode glyphs. putText() (including overloaded variants with custom fonts) will interpret input string as UTF8, so it will be easy-to-use and compatible with the current putText().

These are good news. Maybe not that good news is that it will be proposed for the "next" branch (to be created), so officially it will become available in OpenCV 5.0.

[SteveKrebs]

Well, good to hear it is at least it is a feature on the roadmap (I would sure like to see it in 4.4+ as well). Maybe it should allow to also keep the Hershey fonts for backwards comparability as someone may prefer that (I'm ok with any free fonts that support the extended UTF-8 char sets).

[mshabunin]

BTW, utf8/Cyrillic is already supported with COMPLEX font:

opencv/modules/imgproc/src/drawing.cpp

Lines 2210 to 2252 in 77b01de

	inline void readCheck(int &c, int &i, const String &text, int fontFace)
	{
	int leftBoundary = ' ', rightBoundary = 127;
	if(c >= 0x80 && fontFace == FONT_HERSHEY_COMPLEX)
	{
	if(c == 0xD0 && (uchar)text[i + 1] >= 0x90 && (uchar)text[i + 1] <= 0xBF)
	{
	c = (uchar)text[++i] - 17;
	leftBoundary = 127;
	rightBoundary = 175;
	}
	else if(c == 0xD1 && (uchar)text[i + 1] >= 0x80 && (uchar)text[i + 1] <= 0x8F)
	{
	c = (uchar)text[++i] + 47;
	leftBoundary = 175;
	rightBoundary = 191;
	}
	else
	{
	if(c >= 0xC0 && text[i+1] != 0) //2 bytes utf
	i++;
	if(c >= 0xE0 && text[i+1] != 0) //3 bytes utf
	i++;
	if(c >= 0xF0 && text[i+1] != 0) //4 bytes utf
	i++;
	if(c >= 0xF8 && text[i+1] != 0) //5 bytes utf
	i++;
	if(c >= 0xFC && text[i+1] != 0) //6 bytes utf
	i++;
	c = '?';
	}
	}
	if(c >= rightBoundary || c < leftBoundary)
	c = '?';
	}

[asmorkalov]

Done in #18760