Object Detection In Deep Sky

Object detection in the night sky is an astronomical project aimed at improving our techniques to identify celestial objects and open doors to new discoveries. This project is aimed at detecting celestial objects in the sky using a wide field image of a constellation. Taking advantage of image proce

Project Title

Object Detection In Deep Sky

Project Area of Specialization

Artificial Intelligence

Project Summary

Object detection in the night sky is an astronomical project aimed at improving our techniques to identify celestial objects and open doors to new discoveries. This project is aimed at detecting celestial objects in the sky using a wide field image of a constellation. Taking advantage of image processing techniques, we propose a constellation algorithm that detects constellations directly from a wide-field image of the night sky. Since the constellations in the sky have fixed patterns and neighborhood relationships, it is probable to apply specific template matching techniques to match the constellation templates with the real star patterns on the image. The algorithm will be further used to compare constellation (image taken in some future time period) with the previous constellation image using a comparison algorithm to find any celestial objects that have passed through it. This fulfills the aim of our project and would help us in achieving our goal of detecting objects that pass through a constellation as well as discovering objects in space that have yet not been seen. There are 88 constellations is the sky, out of which 12 are renowned namely; Aries, Taurus, Gemini, Cancer, Leo, Virgo, Libra, Scorpio, Sagittarius, Capricorn, Aquarius and Pisces. These constellations are well known because the sun revolves behind them and thus, they are widely used as the Zodiac signs. When the Sun is at the position of Leo, it is the Leo time period and when behind the Pisces constellation, it is the period of Pisces and so on. Out of the huge number of constellations we are choosing the Ursa Minor Constellation in our project since it is always visible from the ground area of the geographical coordinates we live in.

Project Objectives

The main objective of Object Detection in Deep-Sky is mapping stars into a constellation by understanding its dynamics and to identify objects that pass a certain constellation. This object could be an already discovered object or it could be a new discovery. This would catalyze the on-going astronomical research by automating the space detection techniques hence, removing any human constraints that could occur because of human needs or repetitive work. However, with the main objective, there are certain sub-objectives too. Pakistan has always had a minimal contribution internationally so implementing this project would be a great International contribution. The sheer volume of astronomical data, which is increasing exponentially as well as human inefficiency in repetitive task requires an automated approach to increase our throughput and the space of discoveries. It demands Artificial Intelligence to step up in the Astronomical domain.

Project Implementation Method

Many of the pattern recognition algorithms have been developed for not only stars and constellations, but also for detecting several celestial objects such as planets, asteroids etc. Ming Jiang, Yi-Zheng Ye proposed a star pattern recognition algorithm, which divided the celestial sphere into square areas based on bright stars and then selecting subareas for star pattern recognition, which had small angular separation of stars. Clark S. Lindsey and Thomas Lindblad proposed a method inspired by neural networks techniques by using the histograms of distances of stars from an unknown star andcomparing the feature vectors from the histogram, which although was quite efficient in the case of bright noise, the downside of this approach was it required special hardware to perform effectively as it was quite slow on serial processors because of the comparison of large number of prototype vectors.In this project, our focus would be mainly on the Ursa Minor Constellation. Ursa Minor is colloquially known in the US as the Little Dipper because its seven brightest stars seem to form the shape of a dipper. Ursa Minor is also notable for marking the location of the north celestial pole, as it is home to Polaris, the North Star, which is located at the end of the dipper’s handle. This means that as the Earth rotates, Polaris appears to remain stationary in the sky and all the other stars rotate around it. Hence, Ursa Minor is visible throughout the year and the ideal constellation that could be used for image comparison.

Benefits of the Project

There is one field of science, often overlooked in discussions about AI that can offer valuable direction when trying to determine the purpose of AI, and that is the science of Astronomy. Astronomy studies the Universe and the Universe presents an infinite challenge to AI. If we create an AI for the purpose of cooperating with us in exploring the Universe, then we will develop a partnership that will lead us to other worlds we could possibly inhabit. When scientists discover a new star or galaxy, they typically rely on information gleaned from academic papers, catalogs and other existing information to classify their discovery. This may be typical, but it's not as efficient as it could be. Therefore, finding them has been slow and tedious work. That’s where artificial intelligence comes in. Astronomers no longer keep lonely vigils on cloudless nights, tracking the movement of individual planets; instead, they use sophisticated machinery that guzzles up portions of the sky in gulps of data unimaginable to early scientists. Better telescopes and better data storage means there’s more than ever to analyze. This wave of AI astronomers isn’t just thinking how this technology can sort data. They’re exploring what could be an entirely new mode of scientific discovery, where artificial intelligence maps out the parts of the Universe we’ve never even seen. To conclude, artificial intelligence could be used to create information, filling in blind spots in our observations of the Universe and performing a sort of scientific alchemy,helping us turn old knowledge into new. And we’d be able to explore space like never, without even leaving Earth.

Technical Details of Final Deliverable

1. Image Pre-processing:

The processing and filtering of images is important to remove any type of noise and nonstellar objects that may affect the accuracy in detecting the constellation in the deep sky image. To detect all possible constellations in the image, the bright as well as the dim stars will be preserved in the image after the deep sky image has been filtered properly.

2. Constellation Models:

A database of constellation models will be used that will increase the accuracy of detecting the constellations in the sky image. With respect to each model, the constellation neighbors will also identified such that if any deep sky image has more than one constellation present, then majority of them should be identified. This model database can either be in the form of FITS (Flexible Image Transport System) or simple JPEG, in which case, the models will be pre-processed if required.

3. Pattern Recognition:

The processed deep sky image, will then be passed through the pattern recognition algorithm, which, by matching the templates from the database will detect the constellations from the image by matching the patterns of stars formation in it. As every constellation has fixed number of stars to complete the formation and each constellation has at least one brightest star, these criteria will be matched for accurately detecting the constellation. After detection is complete, the constellations will be plotted on the original image as output.

4. Comparing and Identifying:

The constellation is acquired at another point in time using the above Pattern Recognition technique. Both the constellations are mapped onto one another and differences in both the patterns are identified at the backend using a Pattern Matching algorithm which would allow us to identify any unidentified object that has passed through the constellation.

5. System Design:

The software will be written primarily in Python, as it is one of the most frequently used language for data analysis and image processing. Considering the processing power, we may also use general purpose GPU (GPGPU) to obtain better processing speed pattern recognition algorithms on the deep sky image.

Final Deliverable of the Project

Software System

Core Industry

IT

Other Industries

Others

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Quality Education, Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com