The detection of objects from camera input is very difficult because the external environment cannot be controlled easily. Therefore we use pattern recognition or image processing technologies for detecting objects more accurately.
Imitation of label graphics is seen commonly nowadays hence it becomes necessary for all the organizations to protect their original images. Besides the authenticity of a product also depends upon the fact whether the label graphics are original or not. This can be checked by applying image processing techniques. It basically involves converting the original image data to digital data for the use of the computer.
Image processing techniques let you authenticate or match the images with ease. There can also be a scenario where a person is interested in a particular product and requires to find out the details of the product using its label image, then in such a case the basic requirement that comes into existence is to give best matched results against an image given as a search input from a large set of images.
Considering the above mentioned case, there are some typical techniques which will fulfill the requirement. These are
The SIFT algorithm provides a set of features for image processing that is not affected by many of the problems faced in other methods such as object scaling and rotation. It is invariant to any scaling, rotation or translation of the image.
The SURF algorithm is a robust local feature detector; It can be used in computer vision tasks such as object recognition or 3D reconstruction. The SURF algorithm is faster than the SIFT algorithm. SURF is based on sums of 2D Haar wavelet responses and makes efficient use of integral images.
The ORB algorithm is the most efficient algorithm for image processing. It is an effective method for blob detection. The ORB algorithm uses the FAST (Features from Accelerated Segment Test) and BRIEF (Binary Robust Independent Elementary Features) algorithms.
This project enables the user to match the label image simply by providing the input label image, matching is done from the images present in the database and the best matched image is given as the output. The working is much similar as Google’s image search.