Artificial Intelligence and Machine Learning in Satellite Data Processing and Services

Satellite images are created by transmitting a microwave signal to a target region from a spacecraft or aviation and then analyzing the returned signal. It is a radar image obtained by active sensors that can acquire all weather information regardless of the time of day or weather conditions, unlike typical optical sensors. The noise in the satellite image differs from that found in general photos captured with traditional optical sensors. Many satellite photos struggle from a common problem known as noise. To remove the noise on remotely sensed images, various forms of noise require different approaches. Identifying and denoising noise in remote sensing photographs is a tough task. A characteristic of satellite images is the presence of noise, where this noise often requires preprocessing of the images by a denoising method adapted prior to their compression, storage and transmission. To achieve a considerably higher-grade denoising image of remotely sensed images, wavelet is required and it is reviewed. A novel approach of satellite image enhancement and picture compression technology has been used in this research. When an image is contaminated by noise, it is filtered with a variety of filters before being treated using the DWT and compressed by 2D level space vector quantization, lossy compression technique, which removes noisy bits from the satellite image. Consequently, the picture will arrive ground station in less time.

Mars Orbiter Mission (MOM) is ISRO’s first interplanetary mission. Orbit determination of the spacecraft is always challenging especially for interplanetary missions. It is more difficult when the supported ground stations and the observations are limited. Presently, MOM is being supported by only ISRO’s 32-m antenna for a limited duration. Spacecraft is always facing a lot of disturbances from the sun and various disturbances in Martian orbit. Limited observation data from the ground station resulted in uncertainties in MOM’s orbit. In this paper, efforts are made to tune the MOM’s orbit based on its captured image by its camera and the spacecraft attitude information from the telemetry. Starting from the available coarse orbit information, the orbit parameters were tuned such that the tuned orbit and attitude can produce an image which is exactly matching with the observed one. The paper explains the principle of tuning the parameters using suitable image processing techniques, the System Tool Kit (STK) and MATLAB software. The method described in this paper can be implemented to any interplanetary satellites for orbit tuning as it can be implemented in on-board computer itself. Using present coarse knowledge of orbit, accurate attitude from star sensor, the spacecraft can generate features of the expected reference image. The observed and reference image difference can be used to tune present orbit as done in this paper.

The HEVC is an efficient video compression technique. The proposed study aimed to enhance conventional HEVC by explicitly designing a system that involves the adoption of Graph-based Encoding for compressing HD video frames. However, this research seeks to formulate analytical modeling followed by Multi-Level Optimization (MLO) in encoding to accomplish efficient retention of 8 K resolution in a compatible video frame while performing compression. The experimental setup shows that the proposed method achieves more efficient compression than conventional approaches.

One of the most serious issues that foreign travellers face is traffic congestion. The proliferation of automobiles and a dense population are two major reasons for visitor risk. Reducing visitor dangers is a difficult task, as street injuries account for the bulk of deaths worldwide. Furthermore, many deaths have occurred because of a loss of communication following an accident, even while the victims were still alive. As a result, there is a desire to limit the number of individuals killed or injured on the street. Removing the communication barrier after a crash, particularly while the victims are unconscious, is one strategy to reduce deaths. However, there are numerous options. Main purpose of this tool is to detect car accidents. It is carried out by using a vibration sensor to detect a crash, after which the microcontroller/CPU system receives location data from the GPS and sends it to the GSM, where the system reports on the accident situation on a regular basis. GSM communication platform uses SMS format to send accurate location co-ordinates from GPS to nearby rescue groups such as police departments, hospitals, fire stations, and a few private cell phones. This will allow for quick action to save many people’s lives. In this way, we may break down communication barriers and provide regular contact using Wi-Fi telecom services. This data is likewise stored in a cloud namely ThingSpeak and intimated through satellite.

Sensing of the reflected and emitted radiation from the stressed (biotic and abiotic) crops can be performed using remote and proximal sensing techniques. The hyperspectral dataset’s narrow bands are extremely sensitive and insightful to even minor changes in energy wavelengths. When compared with multispectral datasets, hyperspectral has a much greater potential for analysing crop stress. Secondary hyperspectral data was used in this study to estimate chlorophyll, water content, nitrogen, phosphorus, and potassium in maize. Six machine learning techniques were developed to estimate these parameters using the reflectance values of hyperspectral datasets. The important wavelengths were identified for these parameters using explainable artificial intelligence (XAI) techniques. The optimum wavelengths identified from hyperspectral data were compared with the suitable bands of freely available multispectral satellite imagery (Sentinel-2 MSI, Landsat-7 ETM+, and Landsat-8 OLI). In order to estimate water content, nitrogen, phosphorus, and potassium in maize, the optimum wavelengths identified from hyperspectral data is associated with the SWIR2 band in all the three earth observation datasets. The red-edge2 band of the Sentinel-2 is associated with chlorophyll and nitrogen.

Paddy is a major Kharif crop in Indian Punjab. One of the most critical parameters for ensuring a good harvest and production of paddy is to monitor paddy growth. Satellite data, including optical and radar data can be used to monitor seasonal crops like paddy, which can span a broad region and have a high observation frequency. Cloudy days are common during paddy growing season, making it difficult to obtain cloud-free data. Using radar data, which can penetrate through the clouds, can solve the problem either as a complement to optical data or by itself, to monitor the growth phase of paddy. A study was conducted to detect the three phenological stages (planting, maximum tillering, and harvest) of paddy using VH and VV polarization of Sentinel-1 synthetic aperture radar (SAR) multi-temporal satellite data for Bathinda district of Punjab. Polarization ratio index (RPI = VH/VV) was used to detect the different phenological stages of paddy. The multidate Setninel-2 multi-spectral instrument (MSI) derived normalized difference vegetation index (NDVI) and ground truth locations were used to validate the SAR derived phenological stages. Sentinel-1 data in dual polarization (VH, VV) and both combinations can be used to detect paddy phases such as planting, vegetative, generative, and barren condition after harvesting, according to the findings.

The IoT-based smart irrigation system for the mango orchard focuses on saving of water and getting the better yield, by controlling the amount of water, which is further based on the parameters like weather data from satellites, moisture of the soil, and also plant parameters like age and crop yielding time and many more. In addition to that, the proposed system is going to be robust and smart as it can handle the situations of drought and abundance of water (flood). Here, we use the microcontroller which processes the data available from various sources and gives back the required amount of water needed for the irrigation of mango trees. Also, the whole interface can be accessed via an application. Weather data is available on the Internet which further is taken via remote sensing. Also, the system uses the already available reference values to decide the appropriate value for watering the mango trees which will further help to get a better yield of fruits.

From the beginning of the space era, human beings have launched more than 5500 rockets, many of which have extraordinary kinds of payloads for more than a few of the space missions carried out. These satellites generate utilization but do not maintain utilization; however, they continue to orbit the Earth. This had led to the house’s surroundings being crammed with so many deployed objects that we name house debris. Space particle administration is an elaborate and high-priced problem of growing urgency. Most human beings recognize that the Earth’s orbit is cluttered with debris, but they would possibly not comprehend the full extent. There are over 36,500 particles larger than 10 cm and millions of particles smaller than 10 cm orbiting the Earth. This is a trouble for a couple of reasons. Firstly, as we referred to before, there is simply a ton of accumulated debris particles in orbit around Earth. Secondly, this junk poses a hazard to satellites and house stations. Space particle administration is a challenging hassle due to the fact that there are many one-of-a-kind sorts of particles to think about and it is typically hard to predict when a collision will occur. We used electromagnetic metallic net to collect the space debris with the help of a space craft. This spacecraft will attract small derbies like nuts, bolts, toolboxes, nonmetallic substances, etc. this method will be useful to collect the space debris.

A hybrid renewable energy system (integration solar photovoltaic and doubly fed induction generator) using typhoon HIL real-time simulator is developed. Before an installation of a practical hybrid renewable energy system, the efficiency of the system should verify. This paper proposes an application of a real-time simulator for the hybrid renewable energy system (HRES). The mathematical modelling of doubly fed induction generator (DFIG) and solar photovoltaic (SPV) cells is developed. Moreover, the typhoon hill real-time simulator uses here for validating the accuracy of mathematical modelling of HRES.

Floods due to rainfall and canal breach are the primary concern in Punjab. The intense flooding damages a riparian area’s crop, infrastructure, and transportation. The canal breaching is one of the critical issues which is facing by farmers in this region. The breach caused a flash flood, which damaged crops and caused soil erosion. For rapid response and management, it becomes essential to identify the flooded area and measure the scale of damage. This study uses geospatial techniques for the assessment of the affected area. The study conducted on the canal breach caused in the village Raipura lies in Abohar tehsil of Fazilka district, Punjab. The Cubesat data has been used, which provides high spatial resolution and near real-time imagery. The PCA-based change detection technique is used to assess the affected region. The satellite-based analysis helps in the rapid evaluation of the event over a large area. This study will be helpful for government agencies and insurance companies for immediate relief and to provide compensation to the farmers.

In present-day patterns, knowledge of the video shut circuit for the TV might be fundamental and applicable to the subject of exploration. It is viable with the wide scope of uses, for example, video correspondence and security and surveillance, air terminal, traffic signal, observation exercises at traffic intersections to distinguish blockage, and afterward traffic, and anticipating traffic stream. It is the most unfavorable to the PC is the closet space expected to store this information and recover indistinguishable ones on a case-by-case basis. The proposed work centers around achieving a compelling and productive framework with its insight to keep away from human intercession in distinguishing security dangers. Moving article identification is a staggering impact and powerful exploration point. It becomes applicable to distinguish the real state of the moving item from a given arrangement of video outlines.

Paradigm shift of large satellite to small satellite can be seen in last few decades due to highly increase in applications of small satellites. Due to the revolutionary growth in the technology and availability of Commercial Off the Shelf (COTS) product, it is possible to produce very small but powerful hardware components which are capable of surviving in the dynamic conditions of outer space. However, this is quite challenging task to deploy these small satellites in the harsh space environment. This paper presents various issues and challenges involved in the operation of small satellites and categorization of satellites based on applications is provided.

The mapping and identification of crops and orchards in area is important for forecasting crop yield, evaluating the factors inducing the crop stress, management of resources as well as for the formulation of policy. In Indian Punjab, crop diversification is focused for sustainable agriculture and promoting less water intensive crops. The cultivation of horticultural crops has emerged as one of the viable alternatives for diversification from current paddy-wheat cropping system. Therefore, different orchards were mapped in the Abohar Tehsil of Punjab using Sentinel-2 satellite data (August 2021) which was classified using iso-cluster unsupervised classification technique. Paddy and cotton were differentiated in the near-infrared band (NIR) and orchards from other features using green band. Among different orchards in the area, Kinnow (citrus fruits) was differentiated only from other orchards (like guava, orange, malta and ber) and is cultivated in 13.2% area during 2021. Overall accuracy for classification was 94.1% with the kappa coefficient of 0.87. These results showed that Sentinel-2 can be used for mapping of orchards and monitoring crop diversification.

Movement of earth material, soil, debris, and mud towards downhill is referred as landslide, and this phenomenon occurs due to gravity. It has a very intensive impact on human life including loss of life, damage of buildings, roads and also causes loss of natural resources. There are different kinds of forces that contribute to landslide events, many of them are monitored and observed by remote sensing. Remote sensing provides detail of landslide events, identifies the triggering factors, and monitors the surface activities. Satellite image processing is useful for landslide detection and analysis due to its far-reaching applications. It is also a challenging task due to very high mutability, low resolution, and big database. Last few years satellite remote sensing has become a very important tool for disaster management. Different satellites are designed to collect landslide data, observe landslide prone areas. Focus of this study is to review recent satellite image databases, satellite data analysis and satellite image processing challenges and applications, which gives a baseline for future research on satellite image processing to detect landslides and save human life.

The cities all around the world have observed rapid urbanization which caused the changes in landscape at the expense of productive land. Urban change, which is a result of the urbanization process, has emerged out as a global issue, and it is a major contributor to climate change. One of the measures to mitigate the climate change is increasing green cover in the urban areas. A study was carried out to investigate the effects of green cover on the surface urban heat island in two Indian cities of north-west India using satellite remote sensing: the smart city Ludhiana in Indian Punjab and the city beautiful Chandigarh. Landsat Operational Land Imager (OLI) sensor data (Landsat 8) of May 2021 was used to identify the green cover and built up, and retrieval of land surface temperature (LST) in both the cities. The LST was higher in Ludhiana city than Chandigarh. The LST in Chandigarh was higher in outskirts of the city with less vegetated and unplanned areas than planned core areas of city. In Ludhiana, LST was higher in the south-east direction of the city. The normalized difference vegetation index (NDVI) values were higher in Chandigarh city than Ludhiana city, but NDBI values were higher in Ludhiana city than Chandigarh city. The relationships among LST, NDBI, and NDVI showed a positive correlation with NDBI and negative correlation with NDVI. The urban heat island (UHI) was higher in Ludhiana city than Chandigarh city due to variations in green cover. These results are useful for framing policies by government on the building material and reflective paints along with awareness of people by constructing green roofs to counteract the effect of UHI on human health and sustainable development.

Satellites are widely used for surveillance and natural resource management. Satellites are also used for defense applications as they provide effective and economical solution for communication and surveillance-based applications. The role of satellite is very crucial for applications of disaster management where satellite imagery is of utmost use. The satellite imagery is often affected by the noise due to malfunction of semiconductors and influence of noise during the image communication. This reduces the performance of all the applications based on satellite imagery due to noisy input images. To overcome this issue, amalgamated ROAD-TGM and PCA algorithm is presented in this paper. This algorithm uses the advantages of spatial and transform domain together to improve the accuracy of noise detection. The standard dataset of earth dataset provided by NASA is used to analyze the performance of proposed algorithm. PSNR and SSIM parameters are used to evaluate the performance of proposed algorithm with existing stage of the art algorithms. The proposed method outperforms the existing methods satellite image denoising.

Forests have been an indispensable part of humans and other living beings since the beginning of mankind. Being an abundant source of oxygen, timber, preventing soil erosion, and inhabiting wildlife, flora and fauna, forests play a very important role in maintaining ecological balance. However, the number of fire alerts in forests has increased drastically in recent years. Particularly due to climatic changes, global warming, and man-made mistakes. Such wildfires result in huge economic and environmental damages. Developing an efficient system for monitoring such wildfires is the need of the hour. There have been many methods proposed for this purpose. For hilly terrain applications of forest fire monitoring, wireless networks are one of the most reliable technologies. To design an effective algorithm, data set related to forests such as forest fires, hazard zones, burnt areas are required. The data related to forest fire and burnt area can be gathered accurately using satellite imagery. This paper presents a survey on various satellites using which data is retrieved. The analysis conducted highlights the most preferred satellite opted by researchers for forest-related data sets. This would further be helpful for researchers working towards forest fire detection and other forest-related issues, in choosing the most useful or popular satellite for imagery data sets.

The satellite imagery is one of key source of information gathering associated with environment, natural resource and reconnaissance. As satellite images cover large area and minute details, it is very important to preserve the fine details to obtain highly accurate information. This task is difficult to achieve as images are often affected with noise. So, a preprocessing stage is required which can suppress the level of noise before passing image to applications. This suppression can be achieved by considering each component of data as linear combination of multiple signals. There are many linear transformation techniques available, but ICA is a recently devolved and it is one of the most prominent linear transformations. ICA is a generally utilized for blind source separation technique in which mutual dependence of signal components is minimized. Independent component analysis is applied to numerous applications such as satellite communication, satellite imagery, vibration analysis, speech processing and biomedical signal processing and machinery fault diagnosis. As images are highly influenced by noise due to dual effect of semiconductor aging and image transmission. Hence, it become more critical to remove the noise and reconstruct the natural image by using the image denoising process. The role of ICA in natural image remonstration is expressed in this paper.

This study aimed to assess the feasibility of land sliding factors often used in landslide susceptibility models. For this reason, 12 terrain factors that is, aspect, geology, slope, relative relief, distance to road, distance to drainage, distance to lineament (FFT), profile curvature, stream power index (SPI), topographic wetness index (TWI), land used/land cover (LULC), and soil were identified and analyzed with the help of two approaches, that is weight of evidence and frequency ratio. Important landslide causative factors including landslide inventory data were extracted using remote sensing and ancillary data. Of all the causative factors LULC and aspect have the highest contributing landslide influences according to the data derived approach obtained from landslide inventory. While from the feasibility study also indicates that landslides found in the study area are mostly due to anthropogenic activities. The landslide susceptibility map of the study region was also generated using weight overlay method, and three landslide classes were classified, which are low, moderate, and high zone. From the landslide susceptibility map, it is observed that 10.82% of the area is found in the low class, 77.85% fall under the moderate class, and 11.34% of the area is under the high landslide susceptibility zone. A region with a high susceptibility class should be avoided from further development and planning to prevent further loss of life, property damage, and possible mitigation steps can be taken up.

In Indian Punjab, potato is cultivated during different time periods: early season crop (last week of September–mid-October), mid-season crop (mid-October to mid-November), and late-season crop (second fortnight of January). Time-series mapping of potato is required for characterizing multiple land-use transitions from paddy and other crops to potato. It also helps in accurately discrimination between potato and other cultivated crops such as wheat in Punjab during this time period. Therefore, a study was carried out to classify potato using a time-series analysis of Sentinel-2 imageries in the major potato growing areas (Hoshiarpur, Kapurthala, and Jalandhar districts) of Punjab. The satellite images from October 2021 to February 2022 were classified using an unsupervised classification method. A field survey was carried out in near synchronous with the satellite pass for ground truthing of the classified potato. A difference in spectral reflectance of potato and wheat in near-infrared (NIR) and shortwave infrared (SWIR) bands showed the maximum greenness of potato on 342 and 347 Julian days during 2021. The NIR and SWIR bands were selected as the optimum bands to classify potato from wheat due to the difference in spectral reflectance of these two crops throughout the growth stages. The potato planted in the study area was classified with an accuracy of 86%. During 2021–22, the area under potato was 7.69% (percent of the total cropped area) in the Hoshiarpur district, 12.1% in the Jalandhar district, and 10.1% in the Kapurthala district. These results showed that time series data of Sentinel-2 data is required for accurately identifying the planting window and mapping of potato.

Present research is focused on content-based satellite image retrieval. Boundaries of satellite images have been analyzed using edge detection and compression technique. Edge detection and compression in ant colony optimization and image processing are the subject of current research. The suggested study is centered on the integration of ACO with an edge detection technique and compression. Detection of edges, optimization of ant colonies, and image processing are all topics that have been studied in the past. The difficulties that have been addressed in earlier research are included in the problem statement. According to ant colony optimization requirements, edge detection is necessary when it comes to image processing. Edge detection is also used for image processing, and this is done using the edge detection approach. Edge detection pictures have been studied for contrast, correlation, and entropy as well as for energy, variance, deviation, smoothness, and skewness throughout the feature extraction process.