https://github.com/qsniyg/maxurl
go to section Integrating IMU in your program
or

As mentioned above, userscript.user.js also functions as a node module.

Code:
var maximage = require('./userscript.user.js');

maximage(smallimage, {
  // If set to false, it will return only the URL if there aren't any special properties
  // Recommended to keep true.
  //
  // The only reason this option exists is as a small hack for a helper userscript used to find new rules,
  //  to check if IMU already supports a rule.
  fill_object: true,

  // Maximum amount of times it should be run.
  // Recommended to be at least 5.
  iterations: 200,

  // Whether or not to store to, and use an internal cache for URLs.
  // Set this to "read" if you want to use the cache without storing results to it.
  use_cache: true,

  // Timeout (in seconds) for cache entries in the URL cache
  urlcache_time: 60*60,

  // List of "problems" (such as watermarks or possibly broken image) to exclude.
  //
  // By default, all problems are excluded.
  // You can access the excluded problems through maximage.default_options.exclude_problems
  // By setting it to [], no problems will be excluded.
  //exclude_problems: [],

  // Whether or not to exclude videos
  exclude_videos: false,

  // This will include a "history" of objects found through iterations.
  // Disabling this will only keep the objects found through the last successful iteration.
  include_pastobjs: true,

  // This will try to find the original page for an image, even if it requires extra requests.
  force_page: false,

  // This allows rules that use 3rd-party websites to find larger images
  allow_thirdparty: false,

  // This is useful for implementing a blacklist or whitelist.
  //  If unspecified, it accepts all URLs.
  filter: function(url) {
    return true;
  },

  // Helper function to perform HTTP requests, used for sites like Flickr
  //  The API is expected to be like GM_xmlHTTPRequest's API.
  // An implementation using node's request module can be found in bot.js
  do_request: function(options) {
    // options = {
    //   url: "",
    //   method: "GET",
    //   data: "", // for method: "POST"
    //   overrideMimeType: "", // used to decode alternate charsets
    //   headers: {}, // If a header is null or "", don't include that header
    //   onload: function(resp) {
    //     // resp is expected to be XMLHttpRequest-like object, implementing these fields:
    //     //   finalUrl
    //     //   readyState
    //     //   responseText
    //     //   status
    //   }
    // }
  },

  // Callback
  cb: function(result) {
    if (!result)
      return;

    if (result.length === 1 && result[0].url === smallimage) {
       // No larger image was found
       return;
    }

    for (var i = 0; i < result.length; i++) {
      // Do something with the object
    }
  }
});

The result is a list of objects that contain properties that may be useful in using the returned image(s):

Code:
[{
  // The URL of the image
  url: null,

  // Whether or not this URL is a video
  video: false,

  // Whether it's expected that it will always work or not.
  //  Don't rely on this value if you don't have to
  always_ok: false,

  // Whether or not the URL is likely to work.
  likely_broken: false,

  // Whether or not the server supports a HEAD request.
  can_head: true,

  // HEAD errors that can be ignored
  head_ok_errors: [],

  // Whether or not the server might return the wrong Content-Type header in the HEAD request
  head_wrong_contenttype: false,

  // Whether or not the server might return the wrong Content-Length header in the HEAD request
  head_wrong_contentlength: false,

  // This is used in the return value of the exported function.
  //  If you're using a callback (as shown in the code example above),
  //  this value will always be false
  waiting: false,

  // Whether or not the returned URL is expected to redirect to another URL
  redirects: false,

  // Whether or not the URL is temporary/only works on the current IP (such as a generated download link)
  is_private: false,

  // Whether or not the URL is expected to be the original image stored on the website's servers.
  is_original: false,

  // If this is true, you shouldn't input this URL again into IMU.
  norecurse: false,

  // Whether or not this URL should be used.
  // If true, treat this like a 404
  // If "mask", this image is an overlayed mask
  bad: false,

  // Same as above, but contains a list of objects, e.g.:
  // [{
  //    headers: {"Content-Length": "1000"},
  //    status: 301
  // }]
  // If one of the objects matches the response, it's a bad image.
  // You can use maximage.check_bad_if(bad_if, resp) to check.
  //  (resp is expected to be an XHR-like object)
  bad_if: [],

  // Whether or not this URL is a "fake" URL that was used internally (i.e. if true, don't use this)
  fake: false,

  // Headers required to view the returned URL
  //  If a header is null, don't include that header.
  headers: {},

  // Additional properties that could be useful
  extra: {
    // The original page where this image was hosted
    page: null,

    // The title/caption attached to the image
    caption: null
  },

  // If set, this is a more descriptive filename for the image
  filename: "",

  // A list of problems with this image. Use exclude_problems to exclude images with specific problems
  problems: {
    // If true, the image is likely larger than the one inputted, but it also has a watermark (when the inputted one doesn't)
    watermark: false,

    // If true, the image is likely smaller than the one inputted, but it has no watermark
    smaller: false,

    // If true, the image might be entirely different from the one inputted
    possibly_different: false,

    // If true, the image might be broken (such as GIFs on Tumblr)
    possibly_broken: false
  }
}]