tatan432/computer_vision_basics

Computer Vision Basic Algorithms. SIFT, ORB, Frame Matching, Stitching, Basic Panorama

Folder Description

There are many basic algorithms of computer vision which are either implemented in MATLAB or Python-OpenCV. They can be downloaded or consulted for understanding. Few of the algorithms developed was a part of EE5731 visual computing course (National University of Singapore) requirement fullfillment.

MATLAB Folders:Denoted By PARTX

1. The folder can be saved to any directory.

2. There are six partx_xx folders.

3. Steps to Run the each part - a. Go to each part and extract vlfeat-0.9.21.rar folder. This folder contains the code for SIFT which has been downloaded from - https://www.vlfeat.org/overview/sift.html and is highly cited work.

   b. First matlab working path should be the set to the part directory. For me to run the part1 code I need to set the working directory as -

   E:\<Assignment_directory>\Assignment_1_Sayan_Kumar\Part1
   Similarly to run the code for part2 change the MATLAB working directory to  
   E:\<Assignment_directory>\Assignment_1_Sayan_Kumar\Part2
   

   b. To run each part the following command to run -

    part1: 
    >> image= detect_edge("SOBEL","im01.jpg")
    >> image= detect_edge("GAUSIAN","im01.jpg",sigma,kernel_size) - Example : detect_edge("GAUSIAN","im01.jpg",1,3)
    >> image= detect_edge("HARR","im01.jpg",type,kernel_size) - Example 1:Unit Size Harr Feature Kernel of type 1- >> detect_edge("HARR","im01.jpg",1,1)
       Example 2: Harr Feature Kernel of type  of size 3- detect_edge("HARR","im01.jpg",5,3)
       %%% Unit Size Harr Kernel Types:
       %%% type 1 : (-1,+1) 
       %%% type2: tranpose(-1,+1)
       %%% type 3: (+1,-1,+1)
       %%% type 4: trnaspose(+1,-1,+1)
       %%% type 5: (-1,+1;+1,-1)
       %%% kernel_size will make the kernel bigger by kernel_size times.
   
    part2:
    >> [keypoints, descriptor]=sift_detector("im01.jpg")
   
    Part3:
    >> [H,image]=get_homography("h1.jpg","h2.jpg")  --> h1 to h2 homography
    >> [H,image]=get_homography("h2.jpg","h1.jpg")  --> h2 to h1 homography	
   
    Part4:
    >> [Hm,image]=get_homography_manual("im02.jpg","im01.jpg")  --> im02 to im01 homography and stitching
   
    Part5:
    >> [H12,image]=get_ransac_homography("im01.jpg","im02.jpg")  --> im01 to im02 automatic homography and stitching	
    >> [H21,image]=get_ransac_homography("im01.jpg","im02.jpg")  --> im01 to im02 automatic homography and stitching	
   
    Part6:
    >>[image]=get_basic_panorama("im01.jpg","im02.jpg","im03.jpg");
   
    NOTE: For Part7 to Part10. Download the GCMEX and Road Dataset( https://www.dropbox.com/s/rctkhpw6jh1p22o/Road.zip?dl=0) And add the path of the GCMEX and ROAD Appropriately as shown in the sample code in part1.m in Part7 folder
    part7: 
    >> part1
   
    part8:
    >> part2
   
    Part9:
    >> part3
   
    Part10:
    >> part4
   

   Other Folder :

   ORB_MATCHING: ORB Python-OPENCV Program has been implemented to match two images.

   DATASET:

   Dataset can be downloaded from : https://www.dropbox.com/s/rctkhpw6jh1p22o/Road.zip?dl=0