This level is performed by the API layer KernelAPI. It uses a page-oriented approach with simple output conversion for fast and accurate recognition of the images. This layer provides tools for handling scanners, accessing the image files and various possibilities to adjust the preprocessing, recognizing and post-processing phase of the workflow for creating more accurate and detailed results. It gives direct access to the recognition result for manual administration and processing, or tools for creating some kinds of simple output. The document-level processing provides slower, but more accurate and complex output conversion, for the highest level of layout retention.

The simplest running application is the following (it loads a file test.tif and after recognition, creates a simple text output out.txt - all settings use default values):

    #include "KernelApi.h"
    int main()
    {
        RECERR rc = REC_OK;
        rc = kRecInit(YOUR_COMPANY, YOUR_PRODUCT);
        if (rc != REC_OK && rc != API_INIT_WARN && rc != API_LICENSEVALIDATION_WARN)
            printf("Initialization error occured: %x\n", rc);
        else
        {
            char image[] = "test.tif";
            LPCSTR imgptr[2] = {image, NULL};
            rc = kRecProcessPages(0, "out.txt", imgptr, NULL, NULL, NULL);
            if (rc != REC_OK)
                printf("Processing error occured: %x\n", rc);
            rc = kRecQuit();
            if (rc != REC_OK)
                printf("Quitting error occured: %x\n", rc);
        }
        return kRecGetLastError(NULL, NULL, 0);
    }

Basics
Pages
Recognition
Checking
Output