Template Search in Images using CUDA

This repo contains the implementation of a template search algorithm in images compatible for GPU programming using CUDA.

Solution of Assignment 4 of the course COL380- Introduction to Parallel and Distributed Programming.

Template Search

Given a Data RGB image (call it L) and a small query RGB image (call it Q). The task is to locate the query image Q approximately in the data image L.

Note that the query image Q need not be upright with respect to the data image L. There may be a rotated copy of query image Q in data image L. Thus, a match is specified by the X, Y row-column numbers of the lower-left corner of the query image Q in the data image L and its counter-clockwise rotation in degree of the base of query image Q.

To simplify the problem, the query image is rotated from -45° to +45° in steps of 45°. Image coordinates are (0,0) on the lower left. The required output is a series of <X, Y, degree> triplets. Please note (X, Y) represents row number and column number from the bottom left of Data image L.

Read more here

How to run the Code?

Clone the git repo and enter the Code folder. Execute:

make all

This command compiles all the required files.

./run.sh <path_of_data_image> <path_of_query_image> <threshold_1> <threshold_2> <n as in top n>

This command executes the search program. The threshold 1 and 2 values are required for RMSD calculations and filtering respectively. n is required for outputing top n matched points.