Camera-Calibration & Distortion Correction;Solution

Description

Objectives:

• Learn about camera geometry.

• Learn to calibrate camera using OpenCV functions.

• Learn to obtain and save camera intrinsic and distortion parameters.

• Learn to undistort image from camera.

• Learn to perform pose estimation of a known object.

• You can use either C or C++ version. All functions shown in this document are C++ functions.

Instructions:

• Use OpenCV and a webcam to complete this assignment.

• You will lose 20 points if any of the following requirements is not met:

o Generate a PDF file that includes (with proper headings) one image (Task 1), intrinsic and distortion parameters (Task 2), three difference images (Task 3), rotation and translation matrices (Task 4), intrinsic and distortion parameters (Task 5), one absolute difference image (Task 6), and your explanations.

o Submit your PDF file and source code file(s) in one zip file without the folder or directory. o Use your first name and last name (e.g., justinsmith.zip) as the file name.

• Login to myBYU and submit your work through BYU Learning Suite online submission.

Task 1: Corner Detection 10 points

• Download Camera Calibration Images (bitmap or JPEG format) from BYU Learning Suite.

• Write your code to read in one of those calibration images.

• The input image must first be converted to grayscale using cvtColor() function (CV_RGB2GRAY).

• Use OpenCV function findChessboardCorners() to find chessboard inner corners.

• Use OpenCV function cornerSubPix() to refine corner locations.

• Use OpenCV function drawChessboardCorners() to draw corners.

• Include one output image with corners circled in your PDF file.

• This task is only an intermediate stage of calibration procedure. You don’t have to submit your code.

Task 2: Intrinsic Parameters 20 points

• Write a program to read in all 40 of the calibration images one at a time in a loop.

• In the loop, find chessboard corners for each input image.

• Arrange corner points in the format for calibrateCamera() function. You need to learn how to use vector and vector of vectors if you use C++. A PDF file (Using Vectors and Mat.pdf) can be found in Course Material/Reference on Learning Suite. You can also google or email me if you need help.

• Use OpenCV function calibrateCamera() to calculate the intrinsic and distortion parameters.

• Include the intrinsic (3×3) and distortion (5×1) parameters in your PDF file. (5 point deduction if not in these formats)

• Submit your code for this task.

Task 3: Distortion Correction 20 points

• Write another program to read in your saved intrinsic and distortion parameters from file(s).

• Download the three test images (Far, Close, Turned).

• Use OpenCV function undistort() to correct the distortion of these three images.

• Use OpenCV function absdiff() to compute the absolute difference between the original and undistorted images.

• Include all three absolute difference images in your PDF file.

• Submit your code for this task.

Task 4: Object Pose Estimation 20 points

• Download the “Object with Corners” image to see the known object. You don’t have to process this image. Data are provided.

• Download the data file DataPoints.txt that has 20 image points (for x and y in pixels) and 20 object points (for x, y, z in inches).

• Write a program to read in the image and object points and use your calibration parameters from Task 2.

• Use the C++ version solvePnP() function or C version cvFindExtrinsicCameraParams2() (not preferable) to estimate the object pose (measured by the camera). Convert R and T to 3×3 and 3×1 matrices (5 point deduction if not in these formats).

• Include the output rotation (3×3) and translation (3×1) matrices in your PDF file.

• Submit your code for this task

Task 5: Intrinsic Parameters 15 points

• You can use your real-time acquisition code for Assignment 1 to capture images.

• Use the chessboard for Assignment 1 (blocks must be square) and your code for Task 2 above to calibrate your camera.

• Make sure to change the number of corners entered to the calibration function in your code for Task 2.

• Make sure the chessboard paper is on a planar surface.

• Include the intrinsic (3×3) and distortion (5×1) parameters in your PDF file. (5 point deduction if not in these formats)

• Submit your code for this task.

Task 6: Distortion Correction 15 points

• Repeat Task 3 (including reading calibration parameters from a file) using your own camera.

• Include an image of the absolute difference between the original (captured from your camera) and the undistorted images.

• Submit your code for this task.