Image Recognition Using SIFT

I'm currently doing a face recognition project using SIFT. I have no problem by matching an image with a single image. But, problems come when I want to match an image with a database of image(descriptors). How can i import the database of images and perform matching by using SIFT? The code is here. Thanks in advance.

Answer 1

[Image Analyst]

Put your call to match() in the middle of a for loop like is given in the FAQ: http://matlab.wikia.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F The FAQ will show you how to "import the database of images" assuming that your database is just a collection of files in a folder, and not a true database like Oracle or something.

[Andy Low]

Hi sir, i got a question. The case is, i wanna match an input image(image1) with a database of image(image 2). Currently i loaded a number of images and save it into a database (face_database.dat). The reason i use "image2" is because i wanna loop the images in the database into variable "image2" and perform matching. How can i loop the images in the database into "image2"?

[Image Analyst]

What do you mean by "match"? Can you just subtract the images and look for zeros at every pixel location?

[Andy Low]

The match means matching the keypoints in both images. It is the method of sift, extract feature point and then match with two images. The "match.m" as i provided above will show you.

[Image Analyst]

What do you mean by database? I don't have, nor have I ever used, the Database Toolbox. Do you mean like an Oracle database? If you mean just a folder of images, then it should be pretty obvious how to read in your image from the folder.

Answer 2

[Andy Low]

Thanks so much for the information! I think it will work for me. I will let you know the result after I try it out. Thanks!!