Lab 6: Geometric Correction

Goals: 

Geometric correction is a key image processing in remote sensing. The two main objectives for this lab was to introduce us to the process of collecting and tying ground control points to images. The second process we will be introduced to will be image rectification which will show us how to utilize our geometrically corrected images to realign other images for analysis. Geometric correction is one of the most important aspects within remote sensing and is a tool that ever geographer should know how to accomplish.

Geometric correction allows for images to be placed back into their correct plainometric x,y location and will format pixels into their correct x,y location. Once images have been rectified and geometrically the user now has the ability to extract reliable surface data such as volumetrics. The majority of the correction deals with correcting the overall alignment of the image placing the image into its correct x,y coordinate but it will also correct for any internal errors such as pitch, roll and yaw changes that skewed the image output. 

Methods:

The two main parts of this lab were a) to learn how to collect ground control points and apply image rectification. then b) to apply the new image rectification to register a newer image and geometrically correct the second with a different process. 

Our first task in lab six was to enhance our skills of collecting ground control points and rectifying a image that is geometrically altered. To accomplish his we must first bring in the image that needs to be geometrically corrected in this case we used a image of the Chicago area that was collected in 2000. We then brought in a higher resolution image of the Chicago area to use for collecting and referencing our ground control points. 

From here we opened the control points tool located under multi spectral with the low level of distortion we are able to use the first order polynomial equation and are only required to collect at least three GCP. Once you have opened the control point collector window you need to chose what method of reference type you will use for this particular application. In this case we are using a reference image so we needed to chose the image layer viewer which allows us to place both the distorted image and the corrected image next to one another to collect GCPs. Once the images have been imported the input image is located on the left and the reference image will be positioned on the right. to collect GCPs you will simply click on the create GCP button at the top of the screen that looks like a scope cross heirs. For the first three points you will need to manually place the GCP in the same location on both images. 

You should look for distinctive features to place GCPs such as intersections of highways that will be distinct location or land forms that have defined edges. even corners of buildings are good locations for GCP. After the first three points are collected the rest will synchronize the GCP onto the input image but you will need to manually move them to lower your r-squared value and enhance the overall accuracy of the rectification. A good r-square is .5 or below this will guarantee your rectification to be accurate and able to apply any analysis from that point forward. 

Once you have reached your goal for a root mean square and feel you have enough GCP collected you can now run the resample and rectify the image for geometric correction.  

Part two of the lab was to apply image to image registration this is the process of that takes an already rectified image such as the one that we just corrected and register the next image to correct it to the same x,y coordinates. This part is very similar to part one where we will take a very distorted image of a area in sierra and collect at least 10 GCP with the referenced sierra image to geometrically run a third degree polynomial interpolation however this time we will run a bi linear interpolation for the resampling methods to generate a smoother and more accurate output image of the color contrast. The process used to geometrically correct the image will be the same process as listed before you open control points in the multi-spectral tool bar and input the reference image with the image layer selection. Once at least 10 GCP are collected you can now run the resample and display your out put image.  

Results:

The results of lab six produced two images of both the Chicago area in the 2000 and Sierra. (figure 1) from this lab we where able to geometrically correct two different images with different degree of distortion and experience the skills needed to rectify an image. The lab was able to introduce us to the application and introduce us to the challenges of generating a root mean square under .5. 

In total the lab was very helpful to introducing us to the tools needed for geometric collection with this new skill set I am now able to correctly align any images to their correct plainemetric x,y coordinates. Form there I will be able to analyze a variety of data outputs including volumetrics of soils, or measurements of land features and many more. 

Sources:

Earth resources observation and science center United States Geological Survey

Illinois Geo-spatial Data Clearing House.

Lab 5: LiDAR Applications and basic tools

Basic LiDAR tools and skills 

Goals/Objectives:

The goals for lab five where broken down into three stages. The three parts where separated to introduce all of the students to different tools and skills utilized with LiDAR data sets. The threes sections where

1) Visualizing point cloud data: this step although short was very useful it allowed all the students to be able to understand more clearly what point cloud data is and what the returns do, by looking at the point cloud data within Erdas.

2) Generate LAS dataset and exploring lidar point cloud: Allowed all the students to manage and manipulate the data within ArcMap to calculate the datset, We were able to first generate a LAS dataset, manipulate and calculate the LAS statistics which are used for quality control and assurance. Project the data to correct the orientation of the data. Lastly we explored the different filters and surface displays to understand the capabilities of the different returns. 

3) Generate different surface models with the LAS dataset: Part three introduced the basic tools used with lidar data including DTM ( Digital terrain Model) , DSM (Digital Surface Model), and hillshade tools all three of which generate different output of the terrain found from models of the LAS data.

Methods:

Part one:

Part one of lab three was the only time in this lab we will use Erdas application. In part one all students were able to investigate all of the points that were generated from the LAS data of eau claire county found around the university of Eau Claire. When in Erdas we needed to upload the points to be able to see them within the program. Once uploaded we where then able to see all of the points including all points from first return to last return. (Figure 1)

Figure 1: point cloud generated and displayed in Erdas Imager to show all of us the returns and
the overall layout of our future LAS dataset.

Part Two:

Generateing LAS data and displaying different surface models within ArcMap was a new toll set that many of us students had not been introduced to before this lab. In order to fulfill the task we first needed to upload our LAS data set and to visualize the outputs. To accomplish seeing the data we needed to set up a few parameters prior to bringing in the data. 

First we all needed to set up a LAS data set within one of our personal lab folders to save both the returns and any models we produce from then on. The LAS dataset is similar in function to a geodatabase for vector data. 

Once the LAS data files have been brought in we now have access to all of the statistics that come with it. These statistics are the main use in quality assurance and quality control for it allows you to investigate the min and max of the Z values if you investigate that the points are not close to actual earth then the data may be altered and need to be recollected. 

The second task we needed to complete before we could display the LAS data was to project all of the points in the correct projection. To determine the correct projection we needed to investigate the metadata closer. Once we investigated the metadata we found the correct coordinate system to be assigned to the data was NAVD 1988 US feet. 



Figure 2: shows the four main features that can
be applied during display of Lidar datasets

The four main displays we were showing in our lab included elevation which could be filtered based on different returns. (figure 2). There was also aspect, slope and contour. All of which are capable of applying filters to distinguish which return you want to display.

The last feature we explored in ArcMap was the ability to look at a selected location in a 2-d and 3-d view both of which can be utilized to distinguish features on a map easier. Including bridge structures and even get a better idea of what a building looks like or slope of a ridge. 

Part 3 

The last task of lab five was to generate four different models using our LAS data. We would produce a DSM, DTM hillshade, and an intensity model. Before we can produce these models we needed to determine the spatial resolution that the images would be produced at. This can be accomplished by finding the nominal pulse spacing this is obtained in the LAS statistics and finding the average of the point spacing. The average nominal spacing was found to be about 1.31. 

Now that we have all of our parameters set we now need to run the different tools to generate a DSM and DTM plus a hillshade model for both outputs. 

The steps for conducting this are very simple you just need to use the conversion tool known as LAS to raster, then by changing the internal functions such as value field cell type and void filling you can create wither a DTM or DSM. The difference between the two are within the cell type. For DSM you will select maximum for this will use the last return which in most cases will be the base of earth and the output will be the true surface outline of the area of interest. 

For DTM you will use minimum cell type and binning for the interpolation this will create an output with the first return pulse. The output is more defined and shows more distinct elevation change along with features more pronounced. (figure 3)

Figure 3: Output image generated from the Raster DTM model
the features are more pronounced but do not necessarily show the true earth structure for it utilized the first return.

 

Results 

The results from this lab allowed us to generate five total photos. The first four are both a DSM and DTM of the Eau Claire area mainly located around the University and downtown area. Both of these outputs also had a hill shade model applied to them to intensify the elevation changes. ( FIGURE 4,5)
The last image that was generated was an intensity image which is similar to a DSM but it will darken the contrast and shading of the black and white tones this application amplifies areas of elevation change allowing users to identify changes easier.


Figure 4: DTM hill shade model output.



In total this lab allowed us to enhance our skills of utilizing Lidar data sets. It also taught us the different application that Erdas can apply to Lidar data sets. We where able to identify the three different display methods including (elevation, contour, slope and aspect). We also where introduced to the different returns filters that can be applied these filters allow users to manipulate what return will show on the display option. You can pick from a variety of returns but the main ones are first return which mainly shows the tops of trees or buildings and last return mostly surface elevation.

Figure 5: DSM hill shade model output

In total this lab was very helpful to learning the potential of Lidar application and learning the tools needed for future projects. With this knowledge of lidar data sets I hope to continue to learn to be able to enhance my ability of remote sensing.

Sources:

Lidar point cloud and Tile Index are from Eau Claire county, 2013
Mastering ArcGIS 6th edition data by Margaret Price 2014