Identifying the characters

Aspose.OCR can automatically build a list of characters found in an image and provide less likely alternatives of each character (in descending order of probability). To get the list of characters, use asposeocr_page_characters_choices() or asposeocr_page_characters_choices_from_raw_bytes() functions.

From image

std::string image_path = "source.png";
// allocate memory for a recognized character and 5 alternatives
const size_t len = 4096;
wchar_t buffer[len][6] = { 0 };
RecognitionSettings settings;
size_t res_len = asposeocr_page_characters_choices(image_path.c_str(), buffer, len, settings);
for (size_t i = 0; i < res_len; i++)
{
	// output recognized character
	std::wcout << buffer[i][0] << " - choices: ";
	// all alternatives
	for (size_t j = 1; j < 5; j++)
	{
		std::wcout << buffer[i][j] << " ";
	}
	std::wcout << std::endl;
}
std::wcout << std::endl;

From bytes

unsigned char*** img_raw = <...>;
ImageDescriptor descriptor;
descriptor.image_from_raw_bytes = img_raw;
descriptor.height = 2000;
descriptor.width = 800;
descriptor.channels_size = 3;
// allocate memory for a recognized character and 5 alternatives
const size_t len = 4096;
wchar_t buffer[len][6] = { 0 };
RecognitionSettings settings;
size_t res_len = asposeocr_page_characters_choices_from_raw_bytes(descriptor, buffer, len, settings);
for (size_t i = 0; i < res_len; i++)
{
	// output recognized character
	std::wcout << buffer[i][0] << " - choices: ";
	// all alternatives
	for (size_t j = 1; j < 5; j++)
	{
		std::wcout << buffer[i][j] << " ";
	}
	std::wcout << std::endl;
}
std::wcout << std::endl;

The list of unique characters can later be used for identifying a whitelist to greatly improve the accuracy and performance of recognition.

Alternatives can be used to find out patterns in common recognition errors, such as misidentifying certain characters in a font used in an image, and to take corrective action, such as automatic substitution.