Image Similarity Application Using Ionic, SIFT & KNN

SeeMORE is an Ionic Similarity detection application based on K Nearest Neighbor and Scale Invariant Feature Transform as an Image Feature Extractor.

---

Table Of Contents

• Team Members
• Installation
• Requirements
• Login Page
• Register Page
• Camera Page
• SIFT Feature Extractor
• KNN Algorithm

Team Members

• Salim "Mohammed Salim Khallouki" Sanilogia@gmail.com
• Ouardini "Anas Ouardini" ouardini14@gmail.com
• ElFaroki "Marwane El Faroki" Elfaroki.m.i@gmail.com

Installation

Requirements

• Ionic CLI 5.4.13
• Cordova
• NodeJS v10.15.2
• npm 6.13.4
• JDK 1.8.10
• SDK
• Gradle

Login Page

The First Page of the Application requiring the user to input valid credentials to connect.

Register Page

Represents the Registeration phase of the app making , allowing new users to create valid accounts and connect.

Camera Page

• Shows the main page of the Application , with features allowing to take pictures , upload pictures and crop and modify pictures , in order for them to be processed.

• You can always pick gallery Photos to upload and process.

• Pictures can be edited and croped , allowing for better processing.

SIFT Feature Extraction

• The scale-invariant feature transform (SIFT) is a feature detection algorithm in computer vision to detect and describe local features in images.It locates certain key points and then furnishes them with quantitative information (so-called descriptors) which can for example be used for object recognition. The descriptors are supposed to be invariant against various transformations which might make images look different although they represent the same object,

We will take the following image as an example

By grey scaling the given image and blurring it using A Gaussian convolution , aswell as subjecting it to a sequence of further convolutions with increasing standard diviation , we end up creating a scale space,

By extracting constant features that do no change through the diviations, we can gather a number of invariatiant descriptors.

At last, by removed noise and refining the extrema's, we can extract better points , or we coordinates , that are invariant to translations parallel to the edge direction. these will be represented as a vector of features that can be compared and worked on further.

K Neareast Neighbor

In pattern recognition, the k-nearest neighbors algorithm is a non-parametric method used for classification and regression. It representes an unsupervised learning algorithm that enable classification and sorting , K representes the number elements of a groupings in which the data set will be grouped.

Example :

the following spread is composed of a number of red circles (RC) and green squares(GS). we intend to find out the class of the blue star(BS) , this can be achieved by knn , the k is the nearest neighbors we wish to take vote from , if its 3 , groupings will have 3 in their members given 3 out the 3 members of the stars circle were a red circle , it can be determined , with high confidance , that the star belongs to the (RC) group. the K factor is chosen by running mutiple test examples from 1+ , and comparing results until the varying curve slows down or flatens. the point of diviation could be determined as the optimal value of K. For our current example, given Sift has provided us with a dataset to work , the vectors will be processed bu the knn ,which is provided by ml5.js.