Very simple proof of concept for indexing pdf files iin SharePoint online that are results of scanned content and contain mainly images. This sample uses IronOcr to get the text content of the file.

DocumentTextFromOCR.xml

<Field
    ID="{1012e6c5-2eb5-44cf-9f73-1ed8a584d38b}"
    Name="DocumentTextFromOCR"
    DisplayName="Document Text from OCR"
    Description=""
    StaticName="DocumentTextFromOCR"
    Group="Demo"
    Type="Note"
    NumLines="10"
    UnlimitedLengthInDocumentLibrary="TRUE"
    Hidden="TRUE"
    SourceID="http://schemas.microsoft.com/sharepoint/v3"
/>

Program.cs

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;

using Microsoft.SharePoint.Client;
using OfficeDevPnP.Core;
using IronOcr;

namespace IronPdfTester
{
    class Program
    {
        static string Url = "https://{any_tenant}.sharepoint.com/sites/ocr-tester";
        static string ClientId = "{client_id_guid}";
        static string ClientSecret = "{client_secret_string}";

        static void Main(string[] args)
        {
            using (ClientContext ctx = new AuthenticationManager().GetAppOnlyAuthenticatedContext(Url, ClientId, ClientSecret))
            {
                Console.WriteLine(args[0]);

                Web web = ctx.Web;
                ctx.Load(web);
                ctx.ExecuteQueryRetry();

                Console.WriteLine(web.Title);

                List list = web.Lists.GetByTitle("Documents");
                ListItem item = list.GetItemById(args[0]);
                ctx.Load(item);
                ctx.ExecuteQueryRetry();

                using (System.IO.MemoryStream ms = GetListItemFileStream(ctx, item))
                {
                    string text = MakeOcr(ms);

                    item["DocumentTextFromOCR"] = text;
                    item.Update();
                    ctx.ExecuteQueryRetry();
                }
            }
        }

        static System.IO.MemoryStream GetListItemFileStream(ClientContext ctx, ListItem item)
        {
            System.IO.MemoryStream result = null;

            try
            {
                File file = item.File;
                ctx.Load(file);
                ctx.ExecuteQueryRetry();

                ClientResult<System.IO.Stream> stream = file.OpenBinaryStream();
                ctx.ExecuteQueryRetry();

                result = new System.IO.MemoryStream();
                stream.Value.CopyTo(result);
                result.Position = 0;
            }
            catch (Exception ex)
            {
            }

            return result;
        }
        static string MakeOcr(System.IO.MemoryStream stream)
        {
            var Ocr = new AdvancedOcr()
            {
                CleanBackgroundNoise = false,
                ColorDepth = 4,
                ColorSpace = AdvancedOcr.OcrColorSpace.Color,
                EnhanceContrast = false,
                DetectWhiteTextOnDarkBackgrounds = false,
                RotateAndStraighten = false,
                // Language = IronOcr.Languages.English.OcrLanguagePack,
                EnhanceResolution = false,
                InputImageType = AdvancedOcr.InputTypes.Document,
                ReadBarCodes = true,
                Strategy = AdvancedOcr.OcrStrategy.Fast
            };

            var Results = Ocr.ReadPdf(stream);
            var FullPdfText = Results.Text;

            return FullPdfText;
        }
    }
}