Visual Similarity Duplicate Image Finders for Mac

The world of software continues to advance, and the research centers and private developers continue to push the limits of existing technology in the never-ending quest to design capabilities that fill a very real need.

Take software that finds duplicate files, for example. Software such a Spring Cleaning for Mac from Smith Micro Software, for example, can examine a folder or an entire hard drive for files that are duplicates, even if they have different creation and modification dates or even different names.

Duplicate finding software that searches through a folder on your hard drive should be differentiated from those that work within a specific program, such as iPhoto. For the latter task, you can use a program such as Duplicate Annihilator (Brattoo Propaganda Software) and Tidy Up! by Hyperbolic Software, but these find duplicates based on such characteristics as file name, size, creation date, modification date and so on.

More remarkable, however, is software that finds images that lack an exact match based on visual similarity. There are a few applications for the Mac that can do this. The popular GraphicConverter has a visual similarity search function.

Depending on the version of the software you’re using, you can find matches by choosing Browse Folder from the File menu and selecting the folder you wish to search, then choosing Find, compare and replace from the File menu, and finally choosing Find duplicate files. A dialog box allows you to choose the sensitivity of the search by dragging a slider, but I find sliding it all the way over so as to find the greatest number of matches still results in very few files that aren’t duplicates.

A freeware program with the sole purpose of finding duplicates is PixCompare, whose interface I find to be vastly superior to that of GraphicConverter, because of its large preview windows that give a very good idea of the size and resolution of the matches it finds.

Searches for images based on visual similarity can find images with the same content, even if the two images lack any semblance of a pixel-for-pixel match. Such software can tag two images as duplicates even if one has been heavily cropped, resized in Photoshop, had various filters such as noise removal applied, converted to grayscale or even flipped or rotated.

How does it do this? It looks for areas of similarity in the picture, such as regions of color, texture or shape. Therefore, the finding of matches is not dependent on file size, pixel count or other easily-changed factors.

The drawback is that finding duplicates based on visual similarity is an inexact science, and a large number of false matches will be found, as well as some true matches being overlooked, but fewer duplicates existing on your Mac is better than more. Fortunately, the software tends to put the matches it is more confident about at the front of search results, so you don’t have to look through all the false matches.

Purging duplicate files can save disk space, but in these days of capacious hard drives, the more important benefit is that of saving time. Eliminating duplicate images means that you can spend less time searching through images to find the one you want, as well as avoid editing or tagging the same image more than once. Internet marketers, for example, can search through downloaded folders of PLR or public domain images to weed out annoying duplicates.

Beyond the search for duplicates on one’s hard drive, there are other uses for visual similarity searches. Although visual image software has not quite caught on with the Internet community just yet, the possibilities and potential of visual similarity software for the future are unlimited.

One of the most interesting areas of exploitation in the field of visual image similarity is in the realm of online shopping. Currently, a web or print designer can scan in a copy of a font and a site such as WhatTheFont.com will attempt to identify it, with a good rate of success.

Taking it up a notch, if you see a picture of a necklace that you must have online, such software could locate a similar necklace on one of the online shopping sites. Or perhaps you could snap a picture of your own necklace (or someone else’s) with your iPhone to find matching earrings.

Currently, images must be specifically formatted and indexed in order to make similarity searching possible. This includes such items as jewelry, handbags, watches, shoes and clothing. In the future, other items such as home and garden products may also be added to online databases.

Future uses in the realm of online shopping could include finding a shape that could go nicely with a particular table lamp or finding a sofa that would complement a particular room decor. The possibilities are limitless!