<?xml version="1.0" encoding="utf-8"?>

<!-- Copyright (c) 2007-2010 Nokia Corporation and/or its subsidiary(-ies) All rights reserved. -->

<!-- This component and the accompanying materials are made available under the terms of the License 

"Eclipse Public License v1.0" which accompanies this distribution, 

and is available at the URL "http://www.eclipse.org/legal/epl-v10.html". -->

<!-- Initial Contributors:

    Nokia Corporation - initial contribution.

Contributors: 

-->

<!DOCTYPE concept

  PUBLIC "-//OASIS//DTD DITA Concept//EN" "concept.dtd">

<concept id="GUID-60DBA025-8FA0-5DF2-90D0-744A016998EE" xml:lang="en"><title>Panoramic

Stitching Guide</title><prolog><metadata><keywords/></metadata></prolog><conbody>

<p>This document introduces you to the panoramic stitching functionality. </p>

<section id="GUID-3333CA2A-EF84-5E7B-B9BE-8FF00523897F-GENID-1-10-1-19-1-1-5-1-6-1-11-1-4-1-3-1-4-1-3-1-2-2"><title>Purpose</title> <p>Panoramic

stitching involves combining a collection of images into a single image. </p> </section>

<section><title>Panoramic stitching overview</title> <p>The panorama image

is created by stitching a collection of several separate image into one large

image. The panoramic stitching process is done by, </p> <ul>

<li id="GUID-0DA59FD4-B817-5B51-8E83-F56FC42E1675"><p>image registration </p> </li>

<li id="GUID-74941275-9681-51AB-B1FB-4F0829B47075"><p>cylindrical projection

of images </p> </li>

<li id="GUID-70E87CA6-B8BC-5CE9-A545-FC3782A81A69"><p>image blending. </p> </li>

</ul> <p>The panoramic stitch can be done eihter interactively by using camera

viewfinder, for tracking the camera movements or by stitching pre-captured

images together. </p> <p><b>Introduction

about the panoramic stitching</b> </p> <p>The <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> stitches

a set of images into a single panorama image. The input to <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> is

a collection of high resolution images together with some guidance on the

relative positioning between the images expressed by <xref href="GUID-27BD164A-9283-3BE0-A0BF-A475539D86EA.dita"><apiname>CPanoramaTransform</apiname></xref> instances. </p> <p>For

the panorama creation, the most common use is rendering a full size panorama

image to the device file system. </p> <p>The high level steps to use <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> are

shown below: </p> <ul>

<li id="GUID-EF02A22B-B4B1-5E97-97E7-F09DE26DA93A"><p>To construct <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref>,

use <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-39BE88C3-C617-3AF8-A13D-CEC60AE798B1"><apiname>CImagePanorama::NewL()</apiname></xref> which loads the image panorama

plugin. </p> </li>

<li id="GUID-690408AF-782D-5E67-93BC-238EBCBD0C20"><p>To initialise <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref>,

use <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-B6D24921-92B8-3FD8-B666-91C56E1E7984"><apiname>CImagePanorama::InitialiseL()</apiname></xref>. </p> </li>

<li id="GUID-3E167564-6974-571E-99A0-196A5A18133D"><p>To correct the lens

parameter use <xref href="GUID-90D79D23-97BC-3EE5-98DD-594DF6EFA0EE.dita"><apiname>TPanoramaLens</apiname></xref>. You need to correct the lens

for panorama stitching quality. The lens parameters are camera specific. </p> </li>

<li id="GUID-93217AEB-2AC5-5CCC-B5EE-36E8B99BAA85"><p>To add images to <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref>,

use <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-B6591C30-F990-3D75-B9F3-922D41A5F1A1"><apiname>CImagePanorama::AddFileL()</apiname></xref>,<xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-19FBD32E-9B1F-38B0-963A-B361737818D2"><apiname>CImagePanorama::AddBufferL()</apiname></xref> or <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-66509DE8-2DBA-3FE0-8495-32E4256FA939"><apiname>CImagePanorama::AddImageL()</apiname></xref>, providing a hint on relative positioning or translation of the images through <xref href="GUID-27BD164A-9283-3BE0-A0BF-A475539D86EA.dita"><apiname>CPanoramaTransform</apiname></xref>.

This transform only needs to be approximate. </p> </li>

<li id="GUID-B445F73C-2910-56DE-B03D-2637ECED0963"><p>To make the stitched

image available call <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-A098A70C-050C-3592-BC37-E849A81890F2"><apiname>CImagePanorama::RenderL()</apiname></xref>. </p> </li>

<li id="GUID-B68A2C77-F19A-5999-82C1-F433269DAACB"><p>The <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> instance

can then be deleted. </p> </li>

</ul> <p><b> Introduction

about the CVFTracker</b> </p> <p> <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> is used to

track viewfinder images to aid in capturing a panorama. The usage of <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> are

demonstrated as follows: </p> <ul>

<li id="GUID-3496F712-4FCD-5E89-BFEE-5AA6A22A1DA0"><p>The <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> output

is used for providing feedback and guidance to you, on how to move the camera

while capturing a panorama. </p> </li>

<li id="GUID-BDBC8A06-B2F8-507C-81F7-E2F344A94FEA"><p>The <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> is

used as an aid for the stitching process in the <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref>.

The input to the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> is a rapid flow of raw images

from the viewfinder of the camera. The output is the movement of the camera. </p> </li>

</ul> <p>The high level steps to use the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> are

shown below: </p> <ul>

<li id="GUID-5C25D0C8-49E1-5514-A90C-2B8A026D20CC"><p>To create <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref>,

use the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-AE92A41F-388D-338C-A80E-C0B16DF3BDDC"><apiname>CVFTracker::NewL()</apiname></xref>. The <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-AE92A41F-388D-338C-A80E-C0B16DF3BDDC"><apiname>CVFTracker::NewL()</apiname></xref> is

used to load the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> plugin. </p> </li>

<li id="GUID-ECE3D707-B871-5EC5-A04D-CA2A632BECC6"><p>To initialise <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref>,

use <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-5E441486-3333-3B4F-81F3-3A86AE541CF8"><apiname>CVFTracker::InitialiseL()</apiname></xref>. </p> </li>

<li id="GUID-54FEF93A-68BB-5EFB-8DE4-21F6562DCF11"><p>To reset <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref>,

use <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-AE707DEA-4747-3A1A-8C6F-8163F1DF0584"><apiname>CVFTracker::Reset()</apiname></xref>. </p> </li>

<li id="GUID-32F88242-4EDD-54C7-8D7A-574015A197F4"><p>To perform the image

registration, use <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-AB58BBEB-5EAE-3E2A-A94E-F59334402A2B"><apiname>CVFTracker::RegisterImageL()</apiname></xref>. </p> </li>

<li id="GUID-01DDC476-9596-55A1-B296-13877DB41B67"><p>To destroy the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref>,

use <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-C2B64F7D-04D8-324A-BD84-C588503881BC"><apiname>CVFTracker::~CVFTracker()</apiname></xref>. </p> </li>

</ul> <p><b>Camera

requirements </b> </p> <p>The camera requirements for the <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> and <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> are

as follows, </p> <ul>

<li id="GUID-C83A6919-C7BB-55C9-9D60-80EDAE0D4458"><p>The camera must provide

low resolution images (in the order of 160 by 120 up to 320 by 240 resolution)

as the raw format at a high frame rate to the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> and

high resolution captured images to the <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref>. </p> </li>

<li id="GUID-EF41F660-AAEE-5D31-A014-FD6C39C42C5E"><p>For the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref>,

it is better to have good frame rate than to have high resolution input images. </p> </li>

<li id="GUID-810B06D2-D45F-5E7C-BC62-88AEE0B0A071"><p>The use of large <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> images

can cause a lower frame rate on slow devices, and do not give accurate tracking

for the view finder. </p> </li>

<li id="GUID-DA5C3ED7-550D-5E4E-AD5A-E9119A56D8FB"><p>The optimal size of

images is a tradeoff between the panorama quality and the time for rendering

the panorama. </p> </li>

</ul> </section>

<section><title>How to use Panoramic Stitching</title> <p>This section contains

some code snippets to show, how Panoramic stitching is accessed in several

situations. </p> <p><b>Basic

stitching </b> </p> <p>The basic stitching is demonstrated by a simple example

of stitching four JPEG images into a single JPEG image providing an approximate

translation between each image: </p> <p> </p> <codeblock id="GUID-304B7755-B05B-5BAE-884A-44A0500ABF37" xml:space="preserve">

    TSize imageSize(1200, 1000);

    TDirection direction = EPanoramaDirectionRight;

    //Lens parameters or internal camera characteristics should be available for the 

    //specific camera module in use. Here we use default values.

      TPanoramaLens lens;

    //Create transform. A transform is an approximate offset between each image.

      CPanoramaTransform* panTrans = CPanoramaTransform::NewL();

      CleanupStack::PushL(panTrans);

    //Create panorama object.This stitches the individual images together.

      CImagePanorama* panImage = CImagePanorama::NewL();

      CleanupStack::PushL(panImage);

    //Create panorama engine and set the lens and size

      panImage->InitializeL(imageSize, lens, direction);

    //Each file added is provided with an approximate translation. 

    //In this case there are 4 images.

      panTrans->SetTranslationL(TPoint(0, 0), imageSize);

      panImage->AddFileL(KTestFileName1, *panTrans);

    panTrans->SetTranslationL(TPoint(900, -30), imageSize);

    panImage->AddFileL(KTestFileName2, *panTrans);

    panTrans->SetTranslationL(TPoint(900, 60), imageSize);

    panImage->AddFileL(KTestFileName3, *panTrans);

    panTrans->SetTranslationL(TPoint(400, -30), imageSize);

    panImage->AddFileL(KTestFileName4, *panTrans);

    //The image size can be obtained before rendering (if required)

      TSize size;

      panImage->CurrentImageSizeL(size);

    //view the output image

      panImage->RenderL(KTestBSOutputFileName);

    CleanupStack::PopAndDestroy(2); //panTrans, panImage

</codeblock> <p>The main steps for using <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> are

as follows, </p> <ol id="GUID-D6872BA3-A3DF-566A-83AF-527870CBE997">

<li id="GUID-FB378114-769F-5404-BD20-8886D538191F"><p>create <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref>  </p> </li>

<li id="GUID-D8A28AFC-4D37-59C4-91D9-92F4898F2302"><p>some approximate transform

information for each image pair </p> </li>

<li id="GUID-146D95D7-14DE-57CA-96DA-45959EB895A4"><p>add each images to <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref>  </p> </li>

<li id="GUID-FF479A52-24E3-5D0B-A821-44127DA5B60C"><p>render the output (for

example to the file) </p> </li>

<li id="GUID-0ADBEE3F-AA60-5787-92A0-C421D4EA8731"><p>delete the objects used. </p> </li>

</ol> <p> <b>Note:</b> The following points must be noted by you: </p> <ul>

<li id="GUID-D3EB4190-4B2B-5760-8E6C-D19DAFF439DF"><p>The lens parameters

are not resolution dependant. It is enough to calibrate once for each camera

model and no need to recalibrate for each resolution. For best accuracy, use

the highest resolution possible when calibrating with a calibration tool. </p> <p>The

approximate translations in x co-ordinate and y co-ordinate ( as dx, dy) must

be compliant with the value of direction. </p> </li>

</ul> <p><b>A

camera application example</b> </p> <p>The <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> is

used in a camera application. The example is more complicated than the previous

example. In this example <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> provides the approximate

transform between the captured images. </p> <p><b>Viewfinder

image tracking </b> </p> <p>The <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> generates

panorama images. The application starts the process and then use the camera

for capturing the images. The <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> will give an indication

when to capture the high resolution images, which will later be stitched into

a panorama image. </p> <p>The <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> plays an important

role for getting the camera movement. The high level steps to use the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> in

the <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> for a camera application, are shown

below: </p> <ol id="GUID-B6357C9E-1510-53C8-856B-2C06C7A6A69C">

<li id="GUID-54C90419-08DE-5406-BBF1-37C0B59FCA26"><p>To create the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> object

and load the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> plugin, use <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-AE92A41F-388D-338C-A80E-C0B16DF3BDDC"><apiname>CVFTracker::NewL()</apiname></xref>. </p> </li>

<li id="GUID-D3E16606-ED81-5702-9F4F-9B19754D4768"><p>To create <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> object

and load the <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> plugin, use <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-39BE88C3-C617-3AF8-A13D-CEC60AE798B1"><apiname>CImagePanorama::NewL()</apiname></xref>. </p> </li>

<li id="GUID-E5C6B072-64FE-554F-BAC4-8850E15B841D"><p>To create the <xref href="GUID-27BD164A-9283-3BE0-A0BF-A475539D86EA.dita"><apiname>CPanoramaTransform</apiname></xref> object

and set or get the values, use the <xref href="GUID-27BD164A-9283-3BE0-A0BF-A475539D86EA.dita#GUID-27BD164A-9283-3BE0-A0BF-A475539D86EA/GUID-ADB1D55B-CEA7-3151-B6E2-14F165F8BB9F"><apiname>CPanoramaTransform::NewL()</apiname></xref>. </p> </li>

<li id="GUID-FD68A411-E492-5492-88FD-E7A7307113F7"><p>To create and set the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref>, <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref>,

and <xref href="GUID-27BD164A-9283-3BE0-A0BF-A475539D86EA.dita"><apiname>CPanoramaTransform</apiname></xref> functions for the Image Processor

Adaptation Plug-in objects, use <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-5E441486-3333-3B4F-81F3-3A86AE541CF8"><apiname>CVFTracker::InitialiseL()</apiname></xref> and <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-B6D24921-92B8-3FD8-B666-91C56E1E7984"><apiname>CImagePanorama::InitialiseL()</apiname></xref> respectively.</p> </li>

<li id="GUID-A605F836-474D-55D4-A326-9AC8B0F8F86B"><p>To get a translation

value from a viewfinder image, register an image by using <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-AB58BBEB-5EAE-3E2A-A94E-F59334402A2B"><apiname>CVFTracker::RegisterImageL()</apiname></xref>.

This will return translation information as a CPanoramaTransform object. </p> </li>

<li id="GUID-8E6E35D9-B20A-5BA2-987F-1A3447DEA146"><p>Check the right time

to capture the next image by using <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-945FF7AD-31E5-3C6A-AD89-F1A5FC225B9D"><apiname>CVFTracker::IsTimeToCapture()</apiname></xref>. </p> </li>

<li id="GUID-A490A649-B910-5E1C-B317-EC5627547EF7"><p>Reset the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> for

the next image, using <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita#GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352/GUID-AE707DEA-4747-3A1A-8C6F-8163F1DF0584"><apiname>CVFTracker::Reset()</apiname></xref>. </p> <p>In order

to capture many images to be stitched into a single panorama image, repeat

the steps 5, 6 and 7. </p> </li>

</ol> <fig id="GUID-7F48B2F5-D65B-52EC-8412-F3F56A5EE8E8">

<image href="GUID-470FAFE8-2246-5E91-9F01-6CC95F975E54_d0e304255_href.jpg" placement="inline"/>

</fig> <p><b>Panorama

stitching </b> </p> <p>The <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> helps to decide when

to capture each of the full resolution images using the camera. The images

can then be passed to CImagePanorama and the stitching is done. The high level

steps to do panorama stitching are listed below: </p> <ol id="GUID-763005A3-4098-534F-9CA1-E2F8654C95BD">

<li id="GUID-6866E73F-F1F7-53C8-BA64-8AF448318C53"><p>Capture the next image

using the camera, then add the image to the <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> by

using <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-66509DE8-2DBA-3FE0-8495-32E4256FA939"><apiname>CImagePanorama::AddImageL()</apiname></xref> or <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-201D0607-2D25-3B1D-B449-7A61419BB0C8"><apiname>CImagePanorama::AddFileL</apiname></xref>,

passing in the <xref href="GUID-27BD164A-9283-3BE0-A0BF-A475539D86EA.dita"><apiname>CPanoramaTransform</apiname></xref> object obtained from

the CVFTracker. </p> </li>

<li id="GUID-BC35AEEF-206E-5830-909E-07F645C2E063"><p>When all images have

been captured you can render the stitched image to file, image buffer, CFbsBitmap

or image frame using <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita#GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4/GUID-A098A70C-050C-3592-BC37-E849A81890F2"><apiname>CImagePanorama::RenderL()</apiname></xref>. </p> </li>

</ol> <fig id="GUID-B5F7149F-1FB9-535E-9212-7C9E23B9347B">

<image href="GUID-A8C80AA1-42CF-5866-B223-FCE1AEC9DF63_d0e304305_href.jpg" placement="inline"/>

</fig> <p><b>User Interface</b> </p> <p>Provide a good user interface (UI) application

for the <xref href="GUID-4DBD1F07-42C5-3CD7-B809-A391F315E0D4.dita"><apiname>CImagePanorama</apiname></xref> is crucial. An example of UI design

is described below, which gives you a robust and easy application for generating

panorama images. </p> <p><b>How

to shoot a panorama </b> </p> <p>The steps to be followed for shooting a panorama

images: </p> <ol id="GUID-8FD3D305-BEAF-5414-BD49-5728C2504256">

<li id="GUID-92642922-D0A0-57CD-80A6-03722941E52C"><p>You need to press the

start button of the camera and need to sweep the camera towards the wanted

scene. </p> <p> <b>Note:</b> You must try not to move the mobile phone, but

only rotate it. You see in the below diagram the camera is sweep in a clockwise

direction from left-to-right. The rotation is made around the imagined axis

through mobile (axis of rotation in the image). A panorama image covers approximately

a 100 degree field of view. </p> <fig id="GUID-8562E235-6D10-5097-A888-E27EE6D4E498">

<image href="GUID-7B3578A0-7E2E-508A-A355-A39AEE0307AD_d0e304342_href.jpg" placement="inline"/>

</fig> </li>

<li id="GUID-16CA0CFB-9D17-573C-9D7B-FEDBCFF4044C"><p> <b>Application Start</b>  </p> <p>When

you launch the application the UI appears as shown in the below diagram. </p> <fig id="GUID-4A6082E2-1B7F-5A48-8FC5-7BC2786987B7">

<image href="GUID-0E55E007-913C-56DA-8BEF-7EC00FFCCE51_d0e304358_href.jpg" placement="inline"/>

</fig> <p>Here the screen area shows the viewfinder image. When you sweep

the mobile, several images are captured and then the images are stitched to

create a panorama image. </p> </li>

<li id="GUID-FE8D49DB-26CC-5546-A05D-B46045894313"><p> <b>Capture a panorama

image during tracking</b>  </p> <p>When you choose to capture a panorama image

the UI changes, which is shown in the below diagram. </p> <fig id="GUID-930C8F54-4C13-50AC-B700-9D889FFE9DC7">

<image href="GUID-9AB5481B-C321-56C9-937D-2397AA80DE8E_d0e304377_href.jpg" placement="inline"/>

</fig> <p>In the above diagram you see three different boxes named 1, 2 and

3 which represent the position of the camera. The blue box representation

changes according to the rotation of the mobile phone. </p> <p>The boxes shown

in the above diagram is for illustrative purpose, so that it can be down scaled

for viewfinder image size. The total panorama coverage area box is estimated

by the number of images for capturing and the overlapping area size. </p> <p>The

code snippets below shows the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> tracking. </p> <codeblock id="GUID-73381F9D-53D9-505A-888D-464A83B0A352" xml:space="preserve">

    _LIT(KFileOutput, "c:\\ICLExample\\viewfinder%d.mbm");

    _LIT(KFileCapturedImage, "c:\\ICLExample\\pancapture%d.jpg");

    TInt i = 0;

    TSize imageSize(1200,1000);

 CFbsBitmap* bmp = new (ELeave) CFbsBitmap;

 CleanupStack::PushL(bmp);

    // get 1st viewfinder image from camera into bmp (detail excluded)

             TSize bmpSize = bmp->SizeInPixels();

       CPanoramaTransform* panTrans = CPanoramaTransform::NewL();//create panorama transoform

       CleanupStack::PushL(panTrans);

       CVFTracker* vfTracker = CVFTracker::NewL(); //create VFTracker and load the plugin

       CleanupStack::PushL(vfTracker);

       vfTracker->InitializeL(bmpSize); //Create VFTracker and set size

       CImagePanorama* panImage = CImagePanorama::NewL(); //create CImagePanorama object

       CleanupStack::PushL(panImage);

       TDirection direction = EPanoramaDirectionRight; //assign direction

    // Lens parameters or internal camera characteristics should be available for the 

    // specific camera module in use. Here we use default values.

       TPanoramaLens lens;

       panImage->InitializeL(imageSize, lens, direction); //initialise size, lens, direction and create panorama engine

    // get the first captured image from the camera as a starting point - its name is given in capturedFileName

       TFileName capturedFileName;

    capturedFileName.Format(KFileCapturedImage(),i);

       panImage->AddFileL(capturedFileName, *panTrans); //add the captured image

    do 

        {

        // give the next camera viewfinder image to the tracker    (details ommitted)

           vfTracker->RegisterImageL(*bmp, *panTrans); // register viewfinder image

        // check if we have a good overlap with the previous image

           if(vfTracker->IsTimeToCapture(direction, KPanoramaDefaultOverlap)) 

              {

              // capture the next image from the camera (details ommitted)

              capturedFileName.Format(KFileCapturedImage(),i);

                 panImage->AddFileL(capturedFileName, *panTrans); //add the captured image

                 vfTracker->Reset(); //reset the VFTracker object

              }         

        if ( err != KErrNone )  // some termination condition usually a signal from user

            {

            // no more viewfinder images

            break;

            }

        }

    while (1);    

    panImage->RenderL(KTestVFTOutputFileName); // render the stitched image

    CleanupStack::PopAndDestroy(4,bmp); //panTrans, vfTracker, panImage, bmp

</codeblock> <p>The <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> is passed each viewfinder

images which it checks and determines whether it is right time to capture

the next image. In return it provides translation information <codeph>CPanoramaTransform::GetTranslation()</codeph> which

indicates the shift of the current viewfinder image. </p> <p>The extracted

information is used to draw the blue rectangle on the screen. If it is time

to capture, then it triggers the capturing function and restarts the <xref href="GUID-0D12EA24-3BF5-3BEA-B0EA-CEB89D49D352.dita"><apiname>CVFTracker</apiname></xref> again

for next image. </p> <p> <b>Note:</b> The box representation is used in the

UI, to help you to take better panorama images. In addition a vibration effect

can be used to provide additional feedback on when to capture an image. For

example, when you sweep the mobile phone, it is in tracking mode so it vibrates.

When the mobile phone is ready to capture, the vibration stops to let you

know that it must be static for capturing an accurate second image. Then when

it vibrates again, you must rotate the mobile and until it stop. </p> </li>

<li id="GUID-C7848E42-74B4-56D2-BB7B-2F20A4D30D32"><p> <b>User interface updating

details </b>  </p> <p>The UI design detail is shown in step by step process.

They are as follows: </p> <ul>

<li id="GUID-99D23225-ECF9-5A34-ADAA-4E0A2326D768"><p>The total panorama coverage

area is divided into four different image areas: </p> <fig id="GUID-0808E657-6313-510C-9EB6-F4D49722229B">

<image href="GUID-023D30F0-C1E8-5E09-92AD-C5A7963DCF70_d0e304438_href.jpg" placement="inline"/>

</fig> </li>

<li id="GUID-D06A8B59-81EE-5420-A3AE-FB036B7CE3F4"><p>When you choose to capture

a panorama image the following steps are performed. The first image is captured

at position 0. The UI looks as follows: </p> <fig id="GUID-A17F50EC-3241-56CA-B96A-59515EAF161B">

<image href="GUID-C2D99C88-09C7-55FA-AF95-3F689BA27484_d0e304448_href.jpg" placement="inline"/>

</fig> </li>

<li id="GUID-BF4A2D67-EF8D-5F72-B2C2-9DC79E319023"><p>When you rotate the

mobile phone clockwise, the blue box moves to the right using the translation

information from <xref href="GUID-27BD164A-9283-3BE0-A0BF-A475539D86EA.dita"><apiname>CPanoramaTransform</apiname></xref>. At the same time the

vibrator is set on to give direct feedback for the tracking process. When

the blue box and the grey box (which represent the position to capture the

next image) coincide, the blue box turns into a filled blue box and the vibrator

stops. You then known that the next image is captured. At this moment the

camera must be steady. </p> <p>After second image is captured the blue box

is moving again and vibrator is set-on. The UI looks as follows: </p> <fig id="GUID-C6E583BB-EE47-52EE-BBFB-D7EE37156C12">

<image href="GUID-B985B750-9B42-55ED-B7E4-41B06B44C365_d0e304465_href.jpg" placement="inline"/>

</fig> <p>Again, you rotate the mobile clockwise until the two boxes coincide

and vibrator stops; third image then captured. After this, the UI looks as

follows: </p> <fig id="GUID-FD9D81DA-E675-5C9C-BA6E-4E7DCE6F46D9">

<image href="GUID-9E12593C-38EF-5052-A2E8-0AB25EEFF4D0_d0e304473_href.jpg" placement="inline"/>

</fig> <p>You rotate the mobile phone until the boxes coincide and the fourth

(last) image is captured. </p> <p> <b>Note:</b> If the mobile phone rotation

deviates from chosen (clockwise) direction or translates up and down, the

blue box indicates this, and you must correct the mobile phone for the wrong

movement. </p> <p>At this point the stitched image can be saved or you can

reset the camera, to create another panorama image: </p> <fig id="GUID-1A527BEA-0C50-5910-9525-75336C02B2C9">

<image href="GUID-8E2303E9-DB60-5F38-A4D4-8BF8F7A1EF9F_d0e304490_href.jpg" placement="inline"/>

</fig> </li>

</ul> </li>

</ol> <p><b>Camera

calibration </b> </p> <p>In order to produce good panorama images, the internal

characteristics of the camera is calibrated by the <xref href="GUID-E361A378-76EB-36C9-A949-A5D31D4D1C4C.dita"><apiname>TPanoramaLens()</apiname></xref>.

The lens parameter such as the distance, the focal length, the principal point,

the skew, the width and the height are called the intrinsic parameters. </p> <p>The

camera module supplier should be able to provide this information or there

are a number of tools to get the camera parameters. A free tool that can be

used is the calibration functions in <xref href="http://en.wikipedia.org/wiki/Opencv" scope="external">OpenCV</xref>. Typically these tools need an input of images

of a special reference pattern, for example a checker board like grid, photographed

from many directions. From that the tools automatically generate all the desired

lens parameters. </p> </section>