Joelle Dowling - Revision history

imported>Student221: /* Conclusions */

2020-12-05T01:54:09Z

Conclusions

← Older revision		Revision as of 01:54, 5 December 2020
Line 97:		Line 97:

	== Conclusions ==		== Conclusions ==
	~~Prior to completing this~~ project, I did not have a lot of experience in data analysis. I have learned to identify whether a model is overfitted and whether the data is undersampled. Furthermore, I learned techniques about how to deal with these problems. In the past I was only exposed to singular value decomposition from a mathematical/theoretical perspective, so it was satisfying to use it as an engineering tool. After completing this project, my understanding of how to use ISETCAM to model the image processing pipeline has deepened, by using the sensor capabilities to get real camera data that can be compared with my estimations.		This project provides a good model for predicting RGB values using radiance data. However, there are many ways this work can be built upon. The model could be tested on more datasets and improved from there. Additionally, one could deal more exactly with the problem of overfitting. In this work, a basis was chosen to be less than 8 principal components and to minimize the error when testing the model. However, someone could more closely analyze the data to determine exactly when overfitting starts to occur. The majority of this work was making the model using the full radiance dataset from Measurements1. In the future, someone could explore how accurately the model would be using the data from just one illuminant to train it. Would it be just as accurate when tested on data from a different illuminant as it is when tested on data from same illuminant? Additionally, the project can be modified to predict the CIELAB color image metrics.
	~~In the end,~~ a good model for predicting RGB values ~~was made, at least for the test~~ data ~~used~~. ~~To build upon~~ this work~~, the~~ model could be tested on more datasets and improved from there. Additionally, one could deal more exactly with the problem of overfitting. In this work, a basis was chosen to be less than 8 principal components and to minimize the error when testing the model. However, someone could more closely analyze the data to determine exactly when overfitting starts to occur. The majority of this work was making the model using the full radiance dataset from Measurements1. In the future, someone could explore how accurately the model would be using the data from just one illuminant to train it. Would it be just as accurate when tested on data from a different illuminant as it is when tested on data from same illuminant? Additionally, the project can be modified to predict the CIELAB color image metrics.
			Prior to completing this project, I did not have a lot of experience in data analysis. I have learned to identify whether a model is overfitted and whether the data is undersampled. Furthermore, I learned techniques about how to deal with these problems. Learning to calculate the proportion of the variation in a subset of a dataset was particularly helpful to my understanding of the problems with the results of the simple linear equation. In the past I was only exposed to singular value decomposition from a mathematical/theoretical perspective, so it was satisfying to use it as an engineering tool. I am pleased to say that after completing this project my understanding of how to use ISETCAM to model the image processing pipeline has deepened and I am more confident in my ability to use the program in the future.

	== References ==		== References ==

imported>Student221: /* Appendix: Source Code */

2020-12-05T01:46:29Z

Appendix: Source Code

← Older revision		Revision as of 01:46, 5 December 2020
Line 110:		Line 110:

	== Appendix: Source Code ==		== Appendix: Source Code ==
	~~All the source code and data is saved~~ in this [https://drive.google.com/drive/folders/1ErUfFZ0t-BtggNSgXMv-iNEf13kLEJvD?usp=sharing Google Drive].		Saved in this [https://drive.google.com/drive/folders/1ErUfFZ0t-BtggNSgXMv-iNEf13kLEJvD?usp=sharing Google Drive] is the main source code: ColorCalibration.mlx. There are two folders, Measurements1 and Measurements2, which contain the training and testing data, respectively. Within each of these folders contains a set of Spectral Radiance Measurements and DNG Files.

	== Acknowledgements ==		== Acknowledgements ==
	Thank you to all the instructors of Psych221 for an interesting class this past quarter. A special thank you to Professor Farrell and Professor Wandell for taking all the data needed to complete this project and for their guidance the past couple of weeks.		Thank you to all the instructors of Psych221 for an interesting class this past quarter. A special thank you to Professor Farrell and Professor Wandell for taking all the data needed to complete this project and for their guidance the past couple of weeks.

imported>Student221 at 01:43, 5 December 2020

2020-12-05T01:43:38Z

← Older revision		Revision as of 01:43, 5 December 2020
Line 111:		Line 111:
	== Appendix: Source Code ==		== Appendix: Source Code ==
	All the source code and data is saved in this [https://drive.google.com/drive/folders/1ErUfFZ0t-BtggNSgXMv-iNEf13kLEJvD?usp=sharing Google Drive].		All the source code and data is saved in this [https://drive.google.com/drive/folders/1ErUfFZ0t-BtggNSgXMv-iNEf13kLEJvD?usp=sharing Google Drive].

			== Acknowledgements ==
			Thank you to all the instructors of Psych221 for an interesting class this past quarter. A special thank you to Professor Farrell and Professor Wandell for taking all the data needed to complete this project and for their guidance the past couple of weeks.

imported>Student221: /* Appendix: Source Code */

2020-12-05T01:40:23Z

Appendix: Source Code

← Older revision		Revision as of 01:40, 5 December 2020
Line 110:		Line 110:

	== Appendix: Source Code ==		== Appendix: Source Code ==
	All the source code and data is in this [https://drive.google.com/drive/folders/1ErUfFZ0t-BtggNSgXMv-iNEf13kLEJvD?usp=sharing Google Drive].		All the source code and data is saved in this [https://drive.google.com/drive/folders/1ErUfFZ0t-BtggNSgXMv-iNEf13kLEJvD?usp=sharing Google Drive].

imported>Student221: /* Appendix: Source Code */

2020-12-05T01:40:10Z

Appendix: Source Code

← Older revision		Revision as of 01:40, 5 December 2020
Line 110:		Line 110:

	== Appendix: Source Code ==		== Appendix: Source Code ==
			All the source code and data is in this [https://drive.google.com/drive/folders/1ErUfFZ0t-BtggNSgXMv-iNEf13kLEJvD?usp=sharing Google Drive].

imported>Student221: /* Data and Instrumentation */

2020-12-05T01:37:12Z

Data and Instrumentation

← Older revision		Revision as of 01:37, 5 December 2020
Line 36:		Line 36:
	Since this project creates and tests the model, we have "training" datasets and test datasets. The former are referred to as Measurements1 and the latter, Measurements2. Measurements1 contains radiance data and DNG files for the MCC illuminated by one of three illuminants, Tungsten (A), Cool White Fluorescent (CWF), and Daylight (DAY). Measurements2 contains radiance data and DNG files for the MCC illuminated by a Tungsten light. All the camera images are taken with the same ISO Speed, but have different Exposure Times.		Since this project creates and tests the model, we have "training" datasets and test datasets. The former are referred to as Measurements1 and the latter, Measurements2. Measurements1 contains radiance data and DNG files for the MCC illuminated by one of three illuminants, Tungsten (A), Cool White Fluorescent (CWF), and Daylight (DAY). Measurements2 contains radiance data and DNG files for the MCC illuminated by a Tungsten light. All the camera images are taken with the same ISO Speed, but have different Exposure Times.

	We have to extract the RGB values from the DNG files. For this, we use ISETCAM's sensor functions. First we read the DNG file using <code>sensorDNGRead</code>. Then we open a window that displays the sensor image using, <code>sensorWindow</code>.From there we estimate rectangles where the 24 boxes are on the MCC using the <code>chartCornerpoints</code> function. An example of the sensor window with the rectangles charted is shown in Figure 2. If we are satisfied by the placement of the rectangles, we get the digital values from each square using the function <code>chartRectsData</code>~~. More detail about getting the digital values is in the source code <code>tls_ProjectHelpersSelect.mlx</code> linked in the Appendix~~. We then get the mean and standard deviation of the RGB values for each square. The RGB mean values have to be corrected for Black Level and Exposure Time. This is done by subtracting the Black Level, then dividing the difference by the exposure durations. This is especially important if the exposure durations for the images are different.		We have to extract the RGB values from the DNG files. For this, we use ISETCAM's sensor functions. First we read the DNG file using <code>sensorDNGRead</code>. Then we open a window that displays the sensor image using, <code>sensorWindow</code>.From there we estimate rectangles where the 24 boxes are on the MCC using the <code>chartCornerpoints</code> function. An example of the sensor window with the rectangles charted is shown in Figure 2. If we are satisfied by the placement of the rectangles, we get the digital values from each square using the function <code>chartRectsData</code>. We then get the mean and standard deviation of the RGB values for each square. The RGB mean values have to be corrected for Black Level and Exposure Time. This is done by subtracting the Black Level, then dividing the difference by the exposure durations. This is especially important if the exposure durations for the images are different.
	[[File:SensorWindow.jpg\|center]]		[[File:SensorWindow.jpg\|center]]
	<div style="text-align:center;">Figure 2: Sensor Window example with boxes</div>		<div style="text-align:center;">Figure 2: Sensor Window example with boxes</div>

imported>Student221: /* Best Linear Fit */

2020-12-05T01:20:47Z

Best Linear Fit

← Older revision		Revision as of 01:20, 5 December 2020
Line 69:		Line 69:

	==== Best Linear Fit ====		==== Best Linear Fit ====
	When comparing the observed RGB values with the RGB values predicted by Sony's spectral QE model, there is quite a bit of error, ~~about~~ 0.4. This is shown in Figure 5. Notice that particularly the red channel is off. This discrepancy between Sony's characterization and what is observed could be because of mixing between channels and channel gains being off.		When comparing the observed RGB values with the RGB values predicted by Sony's spectral QE model, there is quite a bit of error, 0.39. This is shown in Figure 5. Notice that particularly the red channel is off. This discrepancy between Sony's characterization and what is observed could be because of mixing between channels and channel gains being off.
	[[File:sony_obs.jpg\|center]]		[[File:sony_obs.jpg\|center]]
	<div style="text-align:center;">Figure 5: Observed and Original Sony Predicted RGB Values</div>		<div style="text-align:center;">Figure 5: Observed and Original Sony Predicted RGB Values</div>
Line 76:		Line 76:
	Now to get the fitted Sony spectral quantum efficiency we just multiply the original Sony curves by the linear transformation L:		Now to get the fitted Sony spectral quantum efficiency we just multiply the original Sony curves by the linear transformation L:
	<div style="text-align:center;">''sonyQE_L = sonyQE * L''</div>		<div style="text-align:center;">''sonyQE_L = sonyQE * L''</div>
	Now, re-predicting the RGB values with the fitted Sony Spectral QE, the error is much lower, ~~about~~ 0.06. The results of this re-prediction are shown in Figure 6.		Now, re-predicting the RGB values with the fitted Sony Spectral QE, the error is much lower, 0.066. The results of this re-prediction are shown in Figure 6.
	[[File:sony_fit.jpg\|center]]		[[File:sony_fit.jpg\|center]]
	<div style="text-align:center;">Figure 6: Observed and Fitted Sony Predicted RGB Values </div>		<div style="text-align:center;">Figure 6: Observed and Fitted Sony Predicted RGB Values </div>

imported>Student221: /* Conclusions */

2020-12-05T01:17:50Z

Conclusions

← Older revision		Revision as of 01:17, 5 December 2020
Line 97:		Line 97:

	== Conclusions ==		== Conclusions ==
	In the end, a good model for predicting RGB values was made~~. To get here~~, ~~I learned about identifying undersampling and principal component analysis (PCA). Additionally, I learned how singular value decomposition is useful~~ for ~~PCA~~.		Prior to completing this project, I did not have a lot of experience in data analysis. I have learned to identify whether a model is overfitted and whether the data is undersampled. Furthermore, I learned techniques about how to deal with these problems. In the past I was only exposed to singular value decomposition from a mathematical/theoretical perspective, so it was satisfying to use it as an engineering tool. After completing this project, my understanding of how to use ISETCAM to model the image processing pipeline has deepened, by using the sensor capabilities to get real camera data that can be compared with my estimations.
	To build upon this work, ~~one could test~~ the model on more datasets ~~to ensure its robustness~~. Additionally, one could deal more exactly with the problem of overfitting. In this work, a basis was chosen to be less than 8 principal components and to minimize the error when testing the model. However, someone could more closely analyze the data to determine exactly when overfitting starts to occur.		In the end, a good model for predicting RGB values was made, at least for the test data used. To build upon this work, the model could be tested on more datasets and improved from there. Additionally, one could deal more exactly with the problem of overfitting. In this work, a basis was chosen to be less than 8 principal components and to minimize the error when testing the model. However, someone could more closely analyze the data to determine exactly when overfitting starts to occur. The majority of this work was making the model using the full radiance dataset from Measurements1. In the future, someone could explore how accurately the model would be using the data from just one illuminant to train it. Would it be just as accurate when tested on data from a different illuminant as it is when tested on data from same illuminant? Additionally, the project can be modified to predict the CIELAB color image metrics.
	The majority of this work was making the model using the full radiance dataset from Measurements1. In the future, someone could explore how accurately the model would be using the data from just one illuminant. Would it be just as accurate when tested on data from a different illuminant as it is when tested on data from same illuminant.

	== References ==		== References ==

imported>Student221: /* RGB Value Comparisons and Error */

2020-12-05T01:05:32Z

RGB Value Comparisons and Error

← Older revision		Revision as of 01:05, 5 December 2020
Line 92:		Line 92:

	==== RGB Value Comparisons and Error ====		==== RGB Value Comparisons and Error ====
	The final results are the outcome when comparing the predicted and observed RGB values for the Measurements2 data, shown in Figure 8. The rms error between the observed RGB values and those predicted by the fitted Sony model is 0.153. The error between the observed RGB values and those predicted by our low-dimensional model is 0.126. So while our estimate is slightly better than the fitted Sony model, both are similar is accuracy.		The final results are the outcome when comparing the predicted and observed RGB values for the Measurements2 data, shown in Figure 8. The rms error between the observed RGB values and those predicted by the fitted Sony model is 0.153. The error between the observed RGB values and those predicted by our low-dimensional model is 0.126. So while our estimate is slightly better than the fitted Sony model, both are similar in accuracy.
	[[File:RGBComp.jpg\|center]]		[[File:RGBComp.jpg\|center]]
	<div style="text-align:center;">Figure 8: Predicted and Observed RGB values with Test Data using a) the fitted Sony model and b) our low-dimensional model</div>		<div style="text-align:center;">Figure 8: Predicted and Observed RGB values with Test Data using a) the fitted Sony model and b) our low-dimensional model</div>

imported>Student221: /* Results */

2020-12-05T01:03:58Z

Results

← Older revision		Revision as of 01:03, 5 December 2020
Line 93:		Line 93:
	==== RGB Value Comparisons and Error ====		==== RGB Value Comparisons and Error ====
	The final results are the outcome when comparing the predicted and observed RGB values for the Measurements2 data, shown in Figure 8. The rms error between the observed RGB values and those predicted by the fitted Sony model is 0.153. The error between the observed RGB values and those predicted by our low-dimensional model is 0.126. So while our estimate is slightly better than the fitted Sony model, both are similar is accuracy.		The final results are the outcome when comparing the predicted and observed RGB values for the Measurements2 data, shown in Figure 8. The rms error between the observed RGB values and those predicted by the fitted Sony model is 0.153. The error between the observed RGB values and those predicted by our low-dimensional model is 0.126. So while our estimate is slightly better than the fitted Sony model, both are similar is accuracy.
	[[File:		[[File:RGBComp.jpg\|center]]
	<div style="text-align:center;">Figure 8: Predicted and Observed RGB values with Test Data using a) the fitted Sony model and b) our low-dimensional model</div>		<div style="text-align:center;">Figure 8: Predicted and Observed RGB values with Test Data using a) the fitted Sony model and b) our low-dimensional model</div>