I2200 Projects 2018 (Due: 12/11/2018)

Transcription

1 1. Project Guideline: I2200 Projects 2018 (Due: 12/11/2018) The project can be either a team project or a single person project. The maximum grade of the project is 100 points and it will be counted toward 40% of the final grade of this course. For the team projects, contributions from each team member should be clearly indicated. A project final report (doc file), presentation (ppt file), and source code must be submitted to TA before midnight of the deadline (12/11/2018). 12/12/2018 and 12/19/2018 are reserved for in-class project presentation and all students are required to attend. 2. Format of Final Report (I attached a format template, please follow this template. The report should between 4 6 pages (cannot be less than 4 pages) Introduction (Problem statement, Motivation) Approach Work Performed (Experiments, software programming, etc.) Results Discussion Reference (cited papers or website) A table listing tasks performed in the project and the percentage contributions by each team member of individual tasks, as well as overall project. Both team members must sign this table. 3. Presentation Schedule 12/12 and 12/19 (Total 15 minutes for each project, 12 minutes presentation, 3 minutes for questions): Project Presentation on 12/12/2018: Project 1: Peter Vertenten Project 2: Jinglun Feng Project 3: Rezwon Bhuiyan Project 4: Dingdong Li Project 5: Md Rashiduzzaman and SM Asraful Alam Project 6: Subash Shrestha Project 7: Sandeep kumar and Azita darvishi Project 8: Tahiya Tazreen and Md. Minhaz Zaman Lasker Project Presentation on 12/19/2018: Project 9: Muhammad Bhatti Project 10: Ching Chan and Tamanna Tanjil Project 11: Munib Ahsan Project 12: Mollie Murray Project 13: Keshav Teeluck Project 14: Michael Ousseinov Project 15: Ashfaq A Khan

2 4. Project Descriptions: Project 1: Image-based document scanner Peter Vertenten Project description: The goal of this project is to create a document scanner, that can use a phone's camera, to take pictures of pages in the document. It would need to enhance contrast, reorient the pixels to account for rotation and sheer, and then ultimately create a pdf from the various snapshots taken. Project 2: Vision-based Inertial Odometry Jinglun Feng Project description: As the front-end in visual-slam, Visual Odometry is mainly applied to estimate mobile robot's position and pose, however, subjected to its high computational complexity and accuracy, properly treating this algorithm remains a challenging problem. One of the solutions is fusion with IMU, a small, inexpensive and very accurate Inertial Measurement Units. In this project, I will implement this method to finish a VIO, which would establish a constraint state between each frame and IMU's data, to track frame feature's more efficiently and in a high speed. The performance of this algorithm would be tested in real environments, involving in Wind Blade or Bridge Detection and indoor environment. Main References: [1] Mourikis A I, Roumeliotis S I. A multi-state constraint Kalman filter for vision-aided inertial navigation[c]//robotics and automation, 2007 IEEE international conference on. IEEE, 2007: [2] Trawny N, Roumeliotis S I. Indirect Kalman filter for 3D attitude estimation[j]. University of Minnesota, Dept. of Comp. Sci. & Eng., Tech. Rep, 2005, 2: [3] Sola J. Quaternion kinematics for the error-state Kalman filter[j]. arxiv preprint arxiv: , Project 3: Land surface classification from satellite images Rezwon Bhuiyan Project description: It is about observing different satellite (Landsat 8, Aqua etc.) images, so that we can measure the water, Ice, bedrock, land or urban's surface area. The idea is to measure the number of pixels that has similar intensity and later find out the overall area. The process could be interesting to make the computer identify or recognize different surface area. However, we can do many different type of research on these satellite images such as cloud detection, rainfall, amount of minerals in water, and fishery in ocean. I can also share my previous research experience with satellite images. Project 4: Image boundary detection applied to Boundary Layer Height (BLH) of Ceilometer data Dingdong Li Project description: Ceilometer is a type of atmospheric lidar measuring cloud ceiling height measurement, boundary layer height and aerosol concentration. Basically, it sends pulse laser to the atmosphere vertically and retrieve the backscatter signal from the

3 atmosphere. Based on scattering theory, different type of particles has different backscatter coefficient. Particularly, the atmosphere has different stratifications or inhomogeneous particle clusters, for example clouds, which can be detected by the backscatter signal of ceilometer. Project 5: Traffic Sign Recognition for Autonomous vehicle Md Rashiduzzaman & SM Asraful Alam Project description: The main objective of this project is to successfully detect traffic signs from a stationary image of a normal street scene. The algorithm should also be able to extract features from the traffic signs and find the best match from predefined templates; also be robust towards variation of the quality of the images. In this project, we want to look into various digital-image-processing algorithms for detection and classification of traffic signs, and compare the performance under different image qualities. Project 6: Extraction of Nutrition Facts from Packaged Food Subash Shrestha Project description: Pictures taken from a phone camera will be analyzed. Any noise on the images will be removed. Text recognition will be applied to extract the texts. Images will be enhanced as needed. If possible I will try to extract nutrition facts from the motion blurred image as well. Project 7: Blurring Face and License Plates for Privacy Protection Sandeep kumar and Azita darvishi Description: For the face detection, we going to use automatic face detector based on robust and efficient face detection algorithm. Skin filter will be used to extract the color and texture information. First of all, I'll be detecting the regions containing human skin in color image. Face detection is done at grey scale level. I've researched for color to grey scale conversion. After skin information extraction, Thersholding and morphology is also used for more information. After detection of skin regions of color image, we'll apply blurring algorithm, only to the face detected regions using matlab. License plate detection: For License plate detection, Automatic license plate detection will be used to extract the vehicle license plate information from an image. Already existing captured images would be used for recognition process. Recognition process will be carried out using character identification based on number plate extraction, splitting characters and template matching. After character recognition, authentification will done. Then I'll be applying blurring algorithm using matlab. Project 8: Application of Optical Character Recognition Tahiya Tazreen and Md. Minhaz Zaman Lasker Project description: Optical Character Recognition (OCR) is a system of converting scanned printed/handwritten image files into its machine-readable text format. It is one of the most interesting and challenging research areas in the field of Image processing. Application of OCR varies from field to field. Through internet we came to know that one widely known

4 OCR application is in banking, where OCR is used to process checks without human involvement. A check can be inserted into a machine, the writing on it is scanned instantly, and the correct amount of money is transferred. On our project, depending upon some characters recognized in the input image of a business card/poster, we will try to develop an application for recognizing the characters of that business card/poster. Project 9: Image-based plant leaf disease detection. Muhammad Bhatti Project description: In this project I will create an algorithm that will allow us to detect the diseases affecting the plants. Since there are many diseases that can affect a plant, I will pick some common diseases for the project. Project 10: Cancer Cell Detection by Digital Image Processing Method Ching Chan and Tamanna Tanjil Project description: For the state-of-art technology, we can apply the digital image processing method to the medical aspect. One of the application of it is to use the related software like Matlab to detection some of the patterns of the behavior of our body. By using the microscopic image from medical device, we can see the cell and tissue directly but it may not be very clear and may not instant interpret some crucial information by looking at it directly. For this project, we target to use the software to recognize some microscopic image that may contain the cancer cells and we will also try to enhance image and provide certain information by emphasize some particular cancer cells. Project 11: Object Distance Measurement Using a Single Camera Munib Ahsan Project description: This project will demonstrate how to measure distance using a single camera. While most autonomous/semi-autonomous cars perform distance measurement using data collected from LiDAR units, radar sensors, and ultra-sonic sensors, but what happens when those equipment fail? These cars are equipped with several cameras for object recognition and detection, but not primarily used for distance measurement. The method proposed in this project will act as a secondary distance measurement system when primary system fails. The project will use real time video stream collected from a single camera and use image processing techniques to be able to detect and measure distance on a moving object. Project 12: Color recognition to assist visually impaired or color blinded people Mollie Murray Project description: The goal of this project is to determine several properties of the colors in a given image. These may include the number of colors in the image, the dominant colors in an image, the distribution of colors in an image, or the similarities between the colors in two images. This project could be extended to include features such as pattern recognition, or auditory feedback to the user. This is a useful project because it can be applied to help blind, color blind, and visually impaired people in general to determine the colors of different objects (e.g., clothes, accessories, and home decor), which will allow them to be more independent. Project 13: Human Detection in Images

5 Keshav Teeluck Project description: The aim of this project is to create an algorithm that can detect humans (in an upright standing or walking position) in images that also have various other objects in them (trash can, car, fire hydrant, pole, cat, dog, etc.). The algorithm would have to account for the differences in shape and size that people exhibit. It should also be able to not detect humans in an image where no humans appear (it should not mistake a random object for a human). Project 14: Enhancing Images with Super Resolution Michael Ousseinov Description: To enhance low resolution images and make them higher resolution with deep learning and/or other algorithmic techniques. The input image would be low resolution of a certain size and the output image would be that image the same size but higher resolution. Applications of this would involve improved quality when upsampling an image and improved analysis of lower quality images. Project 15: Image Watermarking Ashfaq A Khan Project description: Image watermarking is a technique to embed some secret information in the image without changing the format of the image. Applications include copy protection, data authentication, indexing and data hiding etc. Please investigate watermarking methods and see its effects when you do image degradation and compression.