I run a site with lots of small images (www.iconfinder.com) and would like to develop a feature that can compare and recognize images. A user should be able to upload an image (icon) and then the site will respond with information about the image if it's in the database.
What is the approach to finding similar (or the same image). I know I can compare md5 of the two images, but I also want be able to find matches if the are scaled.
This is a good start if you are interested in looking at doing it in PHP:
http://www.intelliot.com/blog/2008/03/sorted-directory-listing-image-resizing-comparison-and-similarity-in-php/
There probably aren't a lot of languages LESS suited to this task than PHP. You should really look for an image comparison library with a C compatible API and figure out how to glue that into your PHP application.
Identical images can be checked with an md5sum, but detecting if somebody uploads a scaled image, which displays the same thing as the other is very hard. This requires digital image processing.
An approach is to scale down all images to a certain width (say 100px). Then check a few coordinates for the color. If another image matches a big part (say 80%), it might be the same image.
But if the image is lighter... this won't work.
Related
This question is slightly long, so I'll try to be clear.
On a website I co-develop, I created a drawing application that sends data about the lines drawn on the <canvas> element to the server to create an image and save it at multiple sizes, which works. There is the standard save which is saved to 320x212, with the smaller version at 240x176.
However, some users on my site use a pretty antiquated device/browser for the application. It also has a smaller viewport, so the image is smaller, so server-side I'm currently multiplying the coordinates to compensate when saving the image to the standard size. But this has the side-effect of random unfilled parts of the image showing up, that were filled in on the canvas.
(source: socialcu.be)
(On the canvas for the smaller device's viewport, the entire bottom portion was green, while my method of multiplying the coordinates to compensate caused many unfilled portions to appear on the scaled-up image)
So my first question is, would making the image as a vector image, scaling it up (while still a vector), and then saving it as a raster image fix that issue?
And secondly (if the answer to the first is a yes), then what would be the best way of doing this in PHP? I've heard of Cairo, but the information on it (namely tutorials for the PHP package) is quite lacking. Optimally, is there a way to do this in Imagick, or a tutorial on how to use Cairo?
Cairo is definitely lacking for documentation on the PHP site, so it will be a much more difficult set to learn. I found a couple of links through Google in regard to Cairo. There is a blog post about someone's experience with it here: Getting started with Cairo
Referenced in that post are a couple of walk-thrus by Michael Maclean.
There are quite a few examples of drawing with imagick on the imagick website, and from the looks of the available research online with examples imagick has a better support community (for now). imagick will output a raster image though.
I'm trying to generate 3D isometric views of players' heads, but I'm not sure what kind of support PHP has for this type of operation, or of any external libraries that may be better suited.
Basically I need to take a net like this (here is a diagram showing what each portion is mapped to) and make a 3D head from it. I also need to include the 'head accessory' portions, which should be slightly larger/offset from the actual head.
Does anyone know how I should go about this?
Well first it will be a complex job in my view.
The http://www.minecraftwiki.net/images/0/01/Skinzones.png file you mentioned is flat, but you have to convert that in ISOMETRIC 3D look, so you have to distort the images
For example look at the images below
So you can see that 3D box image is created from the pieces of other images, the logic is to add perspective to the flat images and join them. but as it is 2D we will call it Image Distortion.
Unfortunately GD Library which comes bundled with PHP is not advanced enough to let you do such things.
You have to use some other library like Image Magic and this link is tutorial for using distort functions http://www.imagemagick.org/Usage/distorts/
Second big thing is the processing of the images, you can process the images live but it will consume lots of resources on server, so it is suggested that you use pre processed images, and not process them every time.
To generate the Isometric image you have to write the code your self, and it may need alteration on each image character depending upon the size of the image. But when you have written a code it will be easy.
My Suggestion is to write your own code once, then alter it for every character and save the processed images in a sprite and use them when you add play functionality.
check out this link as well
http://www.fmwconcepts.com/imagemagick/index.php
suppose there is an image on web without watermark. And someone downloads it and makes some edits on it like adding watermark etc etc. Is it possible to write a script in php to compare these two images. Like when I submit these two images to the script, it should be able to output the original image and manipulated image.
I read google's webmaster page which says
Google often finds multiple copies of the same image online. We use many different signals to identify the original source of the image
Blockquote
This is the main concern of my question
One more doubt is will there be any meta tags inside an image. if at all how to read them. Is it possible to edit them. Are there any information(not visual) inside an image which cannot be edited.
Anything within the image can be edited (it is, after all, just a collection of bytes), and it's definitely trivial for someone to add a watermark to an image, or simply change the contrast ever-so-slightly, to make it a very different file from the original. There are several other non-destructive changes that would make image files look completely different to a naive comparison algorithm (e.g., scaling, changing filetypes and compression, changing brightness, rotation, etc.).
Advanced image processing algorithms, however, can still often identify similarities between images that have been manipulated in ways like those above. There are many algorithms to do this, and honestly you could spend thousands of hours trying to roll an algorithm like this yourself. These sorts of algorithms are referred to as "content-based image retrieval."
You might be better off calling into engine that's already been developed to do exactly this. Here are some possibilities:
TinEye has a RESTful API that you can use, described here.
You could scrape the response from Google's Search by Image results using this technique.
You could use any of the number of suggestions within this slightly older StackOverflow post.
Good luck!
Photos taken by digital cameras usually have exif data embedded.
You can get the data with the exif_read_data function in PHP.
As for identifying similar images, here's some useful resources:
TinEye
SO Q on image similarity
The comments on Resig's article
You could submit both images to ImageEdited and see which one has been edited. Even if the exif data's missing, it tells when an image has been created with a program.
I'm trying to figure out via PHP if I can do the following. Imagine I have a two images. The first is an image of a square (50px by 50px). The second image, is that exact same square, only offset by 5px to the left.
I'm wondering if there is a way to dynamically generate all the images in-between to make this a smooth looking image transition (ie in this case generate the other 4 images to make it look like it moved from left to right). This would be a pretty simple situation, however would be applied to more complex images with the exact same premises, essentially using two images to predict the mediums.
Thank you in advance for your help
php would only be a wrapper, think eval() and google it, like http://www.linuxfocus.org/English/September2001/article139.shtml
regards,
/t
I believe the best thing you could hope to do with PHP is to do a 'morph' animation, where you sample the pixels for the first image and the second image, then create a third image consisting of the average color values. Repeating this process you could create as many sub-frames as you would like, but the result would be a blurring / morphing of image A -> image B, and would only be perceived as an animation with the simplest of input images.
More complicated algorithms including edge-detection or hinted-shape tweening could be utilized and implemented, however PHP might not be the best choice for this.
You can dynamically create images with a combination of PHP and ImageMagick.
You can pass in each dimension to ImageMagick, and it will generate an image on the fly for you. Create an image for each pixel you offset, resulting (in this case) in four extra images you can use later for your animation.
I have many files of a same picture in various resolution, suitable for every devices like mobile, pc, psp etc. Now I am trying to display only unique pictures in the page, but I dont know how to. I could have avoided this if I maintained a database at the first place, but I didn't. And I need your help detecting the largest unique pictures.
Install gd2 and lib puzzle in your server.
Lib puzzle is astonishing and easy to play with it. Check this snippet
<?php
# Compute signatures for two images
$cvec1 = puzzle_fill_cvec_from_file('img1.jpg');
$cvec2 = puzzle_fill_cvec_from_file('img2.jpg');
# Compute the distance between both signatures
$d = puzzle_vector_normalized_distance($cvec1, $cvec2);
# Are pictures similar?
if ($d < PUZZLE_CVEC_SIMILARITY_LOWER_THRESHOLD) {
echo "Pictures are looking similar\n";
} else {
echo "Pictures are different, distance=$d\n";
}
# Compress the signatures for database storage
$compress_cvec1 = puzzle_compress_cvec($cvec1);
$compress_cvec2 = puzzle_compress_cvec($cvec2);
Well, even thou there are quite a few algorithms to do that, i believe it would still be faster to do that manually. Download all the images feed them into something like windows live photo gallery or any other software which could match similar images.
This will take you few hours, but implementing image matching algorithm could take far more. After that you could spend extra time on amending your current system to store everything in a DB.
Fix cause of the problem, not it's symptoms.
Firstly, your problem has hardly anything to do with PHP, so I have removed that tag and added more relevant tags.
Smartly doing it will not require NxN comparisions. You can use lots of heuristics, but first I would like to ask you:
Are all the copies of one image exact resize of each other (is there some cropping done - matching cropped images to the original could be more difficult and time consuming)?
Are all images generated (resized) using the same tool?
What about parameters you have used to resize? For example, are all pictures for displaying on PSP in the same resolution?
What is your estimate of how many unique images you have (i.e, how many copies of each picture there might be - on an average)?
Do you have any kind of categorization already done. For example, are all mobile images in separate folder (or of different resolution than the PC images)? This alone could reduce the number of comparisons a lot, even if you do brute force otherwise.
A very top level hint on why you don't need NxN comparisions: you can devise many different approximate hashes (for example, the distribution of high/low frequency jpeg coefficients) and group "potentially" similar images together. This can reduce the number of comparisions required by 10-100 times or even more depending on the quality of heuristic used and the data set. The hashing can even be done on parts of images. 30000 is not a very large number if you use right techniques.
You should check which of the 2 images is the smallest, take the size of that and then compare only the pixels within the rectangle size.