

SKYDOG ATI LLC
Aerial Thermal Imaging
"Better to Look .... Than Wonder" S.Wallace
SAR (Search and Rescue)
Search and rescue is whole encompassing for people, pets, fawns etc. Sky Dog ATI has searched for fawns during mowing season (see below), to horses, cats, dogs and even chickens. We use the latest drone thermal technologies and typically are between 250-300' above the ground, so are very non-intrusive. Depending on the conditions, we can cover up to 200 acres / hour.
Services include searches for
-
Pet Recovery (Dogs, Horses, Cattle ...) - Terms of Search
-
Hunters, hikers...
Success Stories









Different Types of SAR Operations.
The fundamental difference between Search and Rescue and Search and Recovery lies in the primary objective. Rescue operations are a race against time. Recovery operations are a meticulous effort to locate a person who is presumed deceased, or to find critical evidence.
Search and Rescue (SAR): The Real-Time Mission
In a Search and Rescue mission, the goal is to locate and save a living person. The entire operation is defined by a sense of urgency and immediacy.
-
Core Focus: The operation is a real-time search, where pilots and observers are actively looking for the missing person via a live video feed from the drone.
-
Guiding Intel: The mission heavily relies on the individual's last seen location and the time they went missing.
-
Primary Technology: Infrared (IR) / thermal imaging is the most critical tool. It allows searchers to detect a person's heat signature against the surrounding environment, making it incredibly effective at night, in dense woods, or under challenging environmental conditions. Drones like the DJI M30T are ideal for their all-weather, 24/7 capabilities.
-
Coordination: Missions are typically coordinated with an Incident Commander (IC) and ground teams (also Recovery). The drone acts as "eyes in the sky," using systematic grid types to cover the search area—whether it's open farmland, a dense wooded area, or a complex coastline—and directs ground personnel to points of interest.
Search and Recovery: The Analytical Mission
When the mission shifts to Search and Recovery, the operational focus moves from immediate life-saving to meticulous mapping and data analysis. The process is similar to a rescue, but the goal is to create a detailed, permanent record of the search area for thorough examination.

A Typical SAR Aerial Mapping Mission Workflow
An aerial mapping mission in a recovery operation is a systematic process designed to efficiently clear large areas and identify clues that may be invisible from the ground.
Step 1: Pre-Flight Planning and Assessment 🗺️
Before any drone takes flight, the team conducts a thorough evaluation of the search area.
-
Define Boundaries: The first task is to establish a clear search grid or boundary. This ensures a methodical search and prevents any areas from being missed.
-
Evaluate Terrain: The topography is analyzed to determine the best flight plan. For missions in hilly or mountainous regions, Terrain Follow is a critical feature that allows the drone to maintain a consistent altitude above the ground, ensuring uniform image resolution.
-
Determine RTK Use: For maximum accuracy, the team decides if Real-Time Kinematic (RTK) positioning is needed. In recovery mapping, RTK is used to create centimeter-accurate maps, which is vital for pinpointing the exact location of discovered clues.
Step 2: Drone and Payload Selection 🚁
The choice of drone and sensor depends entirely on the specific goals of the search.
-
RGB Mapping: To create a high-resolution visual map, an RGB camera is used. A typical flight over a 60-acre area at an altitude of 150 feet takes approximately 20+ minutes and captures around 1,000 images. This detailed map is the foundation for visual analysis. Bysides the old school Zoom and Squint. contours can be generated to any level for different goals.
-
Multispectral Mapping: This sensor is used to find non-obvious clues related to vegetation. It can reveal disturbed earth or hidden paths by highlighting differences in vegetation health (a vegetation index).
-
Thermal Mapping: To detect heat signatures that may indicate decomposition, a thermal camera is the primary tool. These flights are flown from 150 to 300 feet—the lower, the better for thermal resolution—with the camera pointed straight down (NADIR) to get the most accurate temperature readings.
Step 3: Data Offload and Processing
Fly mission, then once the drone lands, the captured images are immediately offloaded and prepared for analysis using two primary workflows, often concurrently.
-
Orthomosaic Stitching: The individual images are fed into photogrammetry software to be stitched together into a single, high-resolution, and geographically accurate map called an orthomosaic. This can be an IR, RGB, or Vegetation Index map.
-
SAR Software Processing: The raw images are loaded into specialized SAR software designed for automated analysis. See Step 5.
Step 4: Manual Analysis ("Zoom and Squint") 👀
After the orthomosaic is processed, it is loaded onto a large 4K monitor for detailed manual review. An experienced analyst systematically pans and zooms across the entire map, a process often called the "zoom and squint." They are looking for anomalies—out-of-place colors, unnatural shapes, or any other clue that doesn't fit the natural landscape. Because every pixel is georeferenced, any anomaly can be located with precise GPS coordinates, allowing ground teams to navigate directly to a point of interest with absolute confidence.
Step 5: Automated Analysis with ADIAT 💻
When using the ADIAT (Automated Drone Image Analysis Tool), the analysis becomes a highly focused and efficient process. Instead of manually searching every pixel, the operator leverages sophisticated algorithms to screen the entire block of imaging. ADIAT’s detection algorithms process every image, searching for anomalies. The software flags these areas of interest and presents them to the analyst for manual review. This transforms the tedious "zoom and squint" of an entire map into a much faster process of verifying a targeted list of high-probability images. Crucially, ADIAT was built to run without requiring internet connectivity, ensuring it can be deployed effectively in any environment.
Orthomosaic Example
Example of an orthomosaic (ortho) created from ~900 images at 150' in 4K and around 45 acres.
Best viewed on a 4K monitor, anything less and imagery will get grainy (data loss).
The ortho is overlayed onto Google Maps (top left).
The side by side comparison (top right) is the ortho vs Google Maps, rough estimate to same altitude. Google image is a touch closer. The zoom on the car is the ortho focused on the car at bottom left, up one from blue car. Details of contents in back window good with imaging at 150'. The zoom on the M30T far exceeds this resolution.
All imaging here is based off the orthomosaic at the bottom left. Accuracy is sub inch, any pixel is georeferenced.



