Kheng Cheng Wai as my supervisor to assist me in this project due to the challenging nature of this project. I would like to participate in this project again if given the chance to re-choose the project topic.
Introduction
Motivation and Problem Statement
The main problem is that the existing software cannot automatically add or mark the logo that best matches photos for the photographers so that the logo can be clearly shown to everyone. Finally, the problem mentioned is important to them, the logo added or tagged to photos can help the photographers and artists to protect their copyright from counterfeiting by any party.
Project Scope and Objectives
- Project Title
- Project Objective
- Project Scope
- Project Innovation / Contribution
In this project many innovation or contributions will be given which can help the users to eliminate the problem. To use the information gain approach to compare and find the best position for the logo to be tagged on photos.
Literature Review
Literature Review on Information Gain
This study focused more on how the information gain can determine the distinction between visual targets and non-target images. The aim of this study is to demonstrate that the visual target can be determined with minimal error probability by using minimal error information gain.
Literature Review on Skin Colour Tone Model
The new proposed algorithm will be based on color space and edge detection, which will be explained later. One of them is the new proposed algorithm, and the other is the Robert Cross Edge Detection algorithm combined with the YCbCr color space.
Literature Review on Feature Point Detection
The experiment result showed that the proposed algorithm has very good accuracy and consistency compared to other feature point detectors. Feature point can play the important role in the image processing as it helps to determine which points in an image are significant. There is one condition that must be met by the mutiframe feature point detection method, that is the condition of repeatability.
After eliminating the false feature point candidates, the best feature points are selected after maximizing the feature. To verify the capabilities of the proposed feature point detection method or algorithm in this study, the experiment was conducted and demonstrated. In this experiment testing, these three feature point detection methods will be tested for their success rate in detecting the number of identical feature points in both original and degraded or rotated frames.
The experiment result showed that the proposed method outperforms the other feature point detection methods, but it can only do so when the images are badly blurred and slightly rotated. But the Kitchen and Rosenfeld's feature point detection had the poorer success rate of detection in all cases.
Literature Review on Wavelet Transform
The proposed algorithm was the coevolutionary genetic algorithm that can be used in wavelet transformation. While the coevolutionary genetic algorithm used in wavelet transformation is closely related to the ESP neuroevolution algorithm. The concept and method described in this study can be regarded as an excellent reference for wavelet transformation in image processing, especially image compression.
Different from the new co-evolutionary genetic algorithm proposed for wavelet transform mentioned in previous study, there is another study that focuses on the compression of the medical images using wavelet transform. So to encounter the mentioned problem, wavelet transform is used for the medical image compression. There are some characteristics of wavelet transform that will make the discrete wavelet transform (DWT) to become one of the most important techniques for image compression.
For this reason, the wavelet transform is used in the proposed medical image compression methods. Although this study focuses more on compressing the medical image, the application of wavelet transform to the new proposed compression algorithm can be referenced and implemented to some extent.
Summary of Literature Review on Information Gain
Summary of Literature Review on Skin Colour Tone Model
Summary of Literature Review on Feature Point Detection
Summary of Literature Review on Wavelet Transform
Review of Existing Software
Use the Move tool in Photoshop (or use the keyboard shortcut "V") to click on the logo and drag it onto the image. Use the Move tool in Photoshop (or use the keyboard shortcut “V”) to resize the logo and position it in the desired position in the image. Through the steps above, Photoshop CS6 can efficiently place the logo in each image automatically.
Therefore, the photographers either have to manually place the logo in images one by one or use the automatic way to do those tasks. In addition, Photoshop CS6 can also create the frame for each image and place the logo at the frame (as shown in Figure 2-14). This can be done using a similar process that places the logo in each image as just shown.
This provides greater flexibility in how the logo can be shaped and placed in an image. It cannot detect the best position for the logo to be placed in an image.
Methodology
Block Diagram
System Methodology
- Calculate Mean and Standard Deviation in Moving Kernel
- Compute Information Loss from Adjacent Kernels
- Find Suitable Location for Logo Placement
Once the mean and standard deviation of each kernel movement is calculated and collected, the system will re-intersect the resized image using the kernel created earlier. When the system traverses the size image, in each current core location it will calculate the information loss of each adjacent core. In the current core location, after the information loss of adjacent cores is settled, the system will identify which adjacent core has the least amount of information loss.
The system will mark this adjacent core and store the related information in another database. After calculating the information loss and marking the adjacent core that has the least information loss at each core location, the system will find a suitable location to place the logo so that the logo is clearly displayed on the image. At each kernel location, the system will determine whether all neighboring kernels mark it with the least loss of information.
The system will then use this data collection of determined core location to find a correct location for the logo in the image. Then the system places the logo at the core location, which is found closest to the corners or sides of the image.
Experiment Setting
- Identification of Background and Foreground Using Kullback-Leibler
- Identification of Human Skin Using HSI Colour Space Model
The experimental system will use the kernel to traverse the image iteratively and in each iteration it will calculate the value differences from the image. When the traversal and calculation is done, the experimental system will collect all calculated values and form the probability distribution for the image. Figure 3-6 shows the C# programming code to create the custom kernel based on the size of the logo.
Once the foreground and background are identified, placement of the logo can be easily done as the logo will be placed on the background in the image. So the experimental system should be expected to identify the logo placement locations as shown in Figure 3-7. At first the process is similar to the previous experiment, the kernel is passing through the image.
This means that if the skin pixels occupy the kernel more than 10% of the pixels in the kernel, that region of the kernel will not be selected as the location for placing the logo in the image. There will be many possible places for the logo to be tagged in the image after passing and calculating.
Actual System and Algorithms Walkthroughs
This time, the system will determine how much information is lost to the surroundings from the core itself. At each core position, the system will calculate the information loss from the current core position to the right, bottom, left, and top core positions as shown in Figure 3-14. When the information loss is calculated relative to the core's surroundings, the system will mark the core position that has the least information loss relative to the core's current position.
In each core position, the information about the tagged position will be stored by the system for further use. After calculating the information loss and identifying the core position that has the least information loss in each core position, the system will find the suitable places to place the logo. These marked positions will be identified and saved by the system as the potential positions to place the logo as shown in Figure 3-16.
The potential position will be ignored by the system from inspection when it is in the skin area. After inspection, the system will place the logo at the potential position closest to the side or corner of the photo image.
Conclusion
By combining both algorithms, the logo can be placed in a place of photo image that can be clearly seen by the people.
Resize and watermark multiple images automatically in Photoshop CS6, 2013 (Video file), Available from:
Available at:
BIS (Hons) Information Systems Engineering A-5 Faculty of Information and Communication Technology (Perak Campus), UTAR. This section contains all information from the Expectation Maximization Algorithm 2015) In the univariate Gaussian distribution, the Kullback-Leibler divergence between the Gaussian distribution with mean and variance and the Gaussian distribution with mean and variance is given by . To show how much information will be lost due to the Gaussian distribution pto Gaussian distribution q, let.
