Infrared image link and quality correction
Minimum input:L1
Main output:Available frames, bad pixel map, NUC parameters, quality status
Infrared image algorithm development and equipment adaptation
Infrared image algorithm technical route and input conditions
The technical route is used to break down enterprise requirements into work packages that can be evaluated, developed and accepted. Confirm the minimum input first, and then determine the algorithm, device adaptation, interface and acceptance method.
20 common technical routes
Relative thermal anomalies compared to reference areas
Minimum input:L0/L1
Main output:Hot/cold spots, relative temperature difference proxies, abnormal areas
Radiation Thermometry and Uncertainty
Minimum input:L2
Main output:Temperature matrix, ROI temperature, compensation parameters and error budget
Object recognition and temperature correlation
Minimum input:L2 or L4
Main output:Objects, part IDs, temperature statistics and alarms
Temperature time series and status trend
Minimum input:L3
Main output:Rates of change, steady state values, change points, trends and health indicators
Visible-infrared registration and fusion
Minimum input:L4
Main output:Registration transformation, fusion map, cross-modal features and confidence
Detection and tracking of people, vehicles and animals
Minimum input:L0/L3/L4
Main output:Categories, boxes/masks, tracks, counts and events
Flame, high temperature and heat source analysis
Minimum input:L1/L2/L3
Main output:Fire spots, heat source contours, spreading trends and review clips
Optical gas imaging
Minimum input:Special MWIR/LWIR L3
Main output:Air plume candidates, trajectories, and duration; quantification requires special calibration
Active thermal imaging non-destructive testing
Minimum input:L5
Main output:Defect contrast, phase, depth candidates and POD
Active NIR Imaging and Biometrics
Minimum input:NIR L1/L3
Main output:Iris/vein/eye/reflective target characteristics
SWIR Differences in water content and materials
Minimum input:SWIR L1/L5
Main output:Moisture Proxy, Material Class, Foreign Objects and Regression Values
Infrared multi/hyperspectral analysis
Minimum input:L5
Main output:Spectral classes, endmembers, abundances, anomalies, and compositional regression
Satellite/UAV infrared remote sensing
Minimum input:L4/L5
Main output:Surface temperatures, indices, change zones, and GIS layers
Heat map 3D mapping and robot perception
Minimum input:L4
Main output:3D thermal texture, target coordinates, risk zones and navigation information
Human/Animal Surface Thermal Physiology Research
Minimum input:Controlled L2/L3
Main output:Symmetry, temperature recovery curves and auxiliary indicators
High temperature process and radiation analysis
Minimum input:MWIR/SWIR L2/L3
Main output:High temperature field, molten pool/furnace wall status and process parameters
Silicon wafer, photovoltaic EL/PL and infrared luminescence
Minimum input:NIR/SWIR L1/L5
Main output:Cracks, failure zones, luminescence uniformity and classification
Rules, Tracking and Business Events
Minimum input:L3/L4
Main output:Intrusion, stay, counting, linkage candidates and event fragments
Data, model, deployment and acceptance governance
Minimum input:arbitrary
Main output:Version, hierarchical indicators, drift, alarm audit and recurrence data
Select data and routes based on target questions
| target question | Suggested data | Typical route | Required input |
|---|---|---|---|
| Just find the relative hot/cold spots | L0/L1 thermal image | T02 | Stable display of parameters or original DN, reference area, working conditions |
| Need to report temperature in Celsius | Radial type L2 | T03 | Camera calibration, emissivity, reflected temperature, distance, environment and window |
| Need to find out which device/component is overheating | L2 or RGB-IR L4 | T04/T06 | Asset annotation, field of view calibration, object and temperature pixel mapping |
| Need early warning or analysis process | L3 | T05/T19 | Timestamp, load/process volume, long-term normal and fault data |
| Need to detect people, vehicles and animals at night | LWIR/NIR L3, if necessary RGB-IR | T06/T07 | Target pixel count, distance, weather, occlusion and product domain annotation |
| Internal delamination/voids need to be identified | Active motivation L5 | T10 | Excitation waveforms, sampling synchronization, material parameters, reference test blocks and true values |
| Need to measure water content/material/composition | SWIR/Hyperspectral L5 | T12/T13 | Matching bands, white plate dark field, sample assay true value and batch |
| Need to identify gas | Special OGI L3/L5 | T09 | Corresponding gas absorption band, background ΔT, wind field, distance and controlled release |
| Need for remote sensing of surface temperature/index | Corrected multi-band L5 | T14 | Atmosphere, emissivity/reflectivity, terrain, phase and ground truth |
| Requires medical or human conclusion | Controlled Radiation Data and Compliance Program | T16/T20 | Intended use, environment/subject protocols, ethics, clinical truth and regulations |
Project verification suggestions
When the camera model, spectral band or data conditions have not yet been determined, the key technical routes can be verified through small-scale samples before the complete development scope is determined.
| priority | Verification items | purpose | Minimal data and true value | Suggested verification indicators |
|---|---|---|---|---|
| P0-01 | Raw frames, temperature matrix and AGC display consistency | Confirm whether the data can measure temperature | Camera SDK, RAW, temperature frame, black body two temperature points | Bit depth/endianness is correct; temperature matrix is consistent with SDK |
| P0-02 | Blackbody multi-temperature radiation calibration and NUC | Establish a basis for temperature measurement | Multiple temperature points and different ambient temperatures within the measuring range | Reports MAE/P95, temperature drift and field of view uniformity |
| P0-03 | Fixed equipment hot spot and reference zone temperature difference | Establish a minimum inspection chain | Normal/abnormal equipment, load, contact temperature | Hot Spot Recall, False Alarm Rate and Temperature Error reaches written thresholds |
| P0-04 | Object recognition associated with ROI temperature | Demonstrate "which part is hotter" | RGB-IR or thermal imaging part annotation | Part Recall, mapping errors and temperature statistics can be reviewed |
| P0-05 | Long-term temperature trends and change points | Prove warning availability | Normal/failure timing, load and maintenance records | Change point latency, false alarms/day, and lead times |
| P0-06 | RGB-IR calibration, synchronization and registration | Establish a dual-light foundation | Synchronous dual light, visible infrared calibration target | Registration P95, synchronization error and failure rate |
| P0-07 | Nighttime human/vehicle/animal detection | Verify target awareness | Target distance layered product domain video | Hierarchical Recall, False Alarms/Hour, and End-Side Latency |
| P0-08 | SWIR Water content/material separability test | Decide if it’s worth modeling | Sample gradient, whiteboard dark field, laboratory true value | Spectral separability and set-out batches RMSE |
| P0-09 | Active Pulse Thermography Reference Block | Verify internal defect routes | Known size/depth defects, excitation and synchronization sequences | POD, minimal defects and depth delamination results |
| P0-10 | Photovoltaic module drone thermal imaging | Verify inspection chain | Irradiance/Wind/View, IR+RGB, IV/EL True Value | Component positioning rate, hot spot recall, geographical error |
| P0-11 | Dedicated OGI controlled gas release | Verify gas route | Gas/flow/wind/distance/background controlled testing | Detection rates and false alarms stratified by condition; quantitative verification separately |
| P0-12 | Data, model and reporting traceability | Prevent results from being reproducible | Raw data, parameters, versions, hashes and blind test sets | Any result can be returned to the original frame, configuration and model version |
Technical route acceptance principles
Each route fixes input, truth values, code/model, parameters, dependencies, hardware and failure conditions respectively. The combined scheme should also verify coordinates, timestamps, units, error status, and degradation paths between modules.
Each acceptance threshold is determined in the project plan based on the target equipment, sample range, working conditions and deployment platform.
Project consultation
xuzhiyang0928@gmail.com
Define the Development Scope with Samples and Device Requirements
Infrared image algorithm technical route and input conditions
Please describe the task object, original picture or video, target indicators, device platform, interface and sample situation.
Phone: 13910119357xuzhiyang0928@gmail.com