Claritas Forensic Workstation

**Clarity you can reproduce. Evidence you can defend.**

---

## Overview

**Claritas** is a forensic image and video analysis workstation for courts, law enforcement, security operations, and accredited labs. It unifies evidence loading, frame analysis, classic enhancement, photogrammetry-style measurement, redaction, and **evidence-chain reporting** in a single **non-destructive filter chain**—every step is traceable, reproducible, and auditable.

Claritas uses **classic digital image processing only** (OpenCV-based, reviewable implementations). There is **no AI, no generative models, and no ONNX inference**. Every enhancement comes from documented methods—deconvolution, multi-frame fusion, contrast recovery, stabilization—not from black-box upscaling. That makes it suitable when you must explain in court **what was done and why it is trustworthy**.

> **Everything is a Filter** — load, select frames, enhance, measure, annotate, export, and report in one chain. Original evidence stays read-only; results are always new files.

**Current release:** Claritas Forensic Workstation **1.0.1** (MVP)

---

## Why Claritas

| Principle | What it means |
|-----------|----------------|
| **Reproducible** | Same input, parameters, and software version produce the same output. Projects (`.clp`) are readable JSON; reopening re-runs the chain from scratch. |
| **Auditable** | Each step logs algorithm ID, parameters, operator, timestamps, and SHA-256 hashes for inputs and outputs. |
| **Defensible** | One-click **HTML + JSON** evidence reports with algorithm notes, hash chain, software version, and device fingerprint. |
| **Non-destructive** | Source files are never overwritten. Exports are new files; original hashes remain in the report. |

---

## Core capabilities

### Evidence import and case projects

- **Images** (PNG, JPG, BMP, TIFF, …), **video** (MP4, AVI, MKV, …), and **image sequences**
- **Proprietary DVR / damaged video** conversion via ffmpeg (common international formats plus widely used regional brands such as Hikvision and Dahua)
- **Milestone XProtect archives** — PQZ, XML, PIC, BLK, and PIC sequence folders (dedicated `milestone_archive` loader and **File → Open Milestone Archive…**)
- **RTSP / HTTP network streams** — record to a local evidence file first, then analyze (**Record from RTSP / Network Stream…** and `rtsp_capture` filter; forensic-friendly, not live-only processing)
- **Case projects** (`.clp`) — filter chain, bookmarks, frame range, case ID, exhibit numbers, operator metadata
- **Read-only import** of Amped FIVE (`.afp`) and VisiTrack (`.vis`) projects for migration and side-by-side review
- **New / Open / Close Project** with save prompts when switching cases

### Non-destructive filter chain (History)

- Filter tree with categories and search; double-click to add steps
- Enable/disable, reorder, and **alternative chain variants** to compare processing paths
- **F5 Complete Processing** — full run from original media, hash commit, evidence chain lock
- Before/after **compare** preview, zoom/pan, and **processing ROI** to limit later filters

### Classic enhancement and restoration

Built for CCTV and body-worn workflows, including:

- **Tone and contrast** — curves, levels, auto-levels, gamma, white balance, dehaze, CLAHE, histogram match
- **Sharpen and denoise** — USM, Laplacian, guided filter, Wiener, TV, homomorphic / Retinex, and related classic tools
- **Deblurring** — motion and defocus, blind deconvolution, Richardson–Lucy, interactive PSF, variable motion deblur
- **Multi-frame** — averaging, temporal smoothing, motion-based frame selection, stabilization and integration, classic / perspective super-resolution (IBP)
- **Geometry** — perspective correction and registration, lens undistort, fisheye unwrap, room unroll

### Measure, annotate, present

- **2D measure** — scale calibration, length, angle, area
- **3D ground measure** — six-point ground calibration, height, ground distance
- **Annotation and privacy** — text/shapes, timestamps, Haar-based face anonymization, region redaction
- **Compare and write** — side-by-side / swipe compare, write image/video/sequence, export readable chain scripts

### Verification and evidence reports

- **File info and SHA-256** — view, copy, verify
- **Evidence precheck** — quality scan and workflow suggestions on load
- **Copy and verify** — forensic copy with hash manifest
- **Export evidence report** (HTML + JSON + signature sidecar), single-report and batch verification
- **Algorithm court notes** (F1) — nature of each filter (enhancement vs reconstruction; no generative claims)

### Workflow and productivity

- **Scene wizard** and built-in templates (night plate, low-light lift, multi-frame SR, DVR destripe, and more)
- Save the current chain as a **custom template**; run **assistant scripts** (`.cas`)
- **UI languages:** English, Simplified Chinese, Traditional Chinese (restart required after switch)
- Bundled **Samples** beside the executable (including Milestone PQZ/XML example cases)

---

Typical scenarios

- License plates, signage, and faces in CCTV — classic enhancement, full evidence report
- Night / low-light surveillance** — curves, CLAHE, multi-frame fusion
- Shake and motion blur** — stabilize, deblur, integrate frames
- Proprietary DVR or corrupt video** — transcode, audit original vs cache paths in project notes
- VMS and IP camera forensics** — Milestone export or RTSP/HTTP capture into the same chain
- Publication copies** — redaction/anonymization separate from the analysis master
- Lab and court packets** — structured reports with parameters, hashes, and algorithm explanations

---

## Who it is for

- Forensic video and image examiners in accredited labs
- Law enforcement technical investigation and video units
- Prosecution, defense, and court-appointed experts who need **explainable** workflows
- Security and quality-assurance video analysts
- Insurers and law firms that require reproducible imaging conclusions—not opaque AI

---

## Product highlights

**Explainable classics, not black-box AI**
No generative repair, AI super-resolution, or synthetic face reconstruction. Every step has a named algorithm and parameters you can cite in reports and testimony.

**Offline-first**
Processing runs locally. No cloud model calls—suitable for sensitive cases and air-gapped labs.

**International ingest, court-ready documentation**
Milestone archives and RTSP/HTTP capture join DVR transcode and standard media loaders. Reports and UI support English and Chinese locales.

**Transparent implementation**
OpenCV-based classic pipeline; filter parameters and chain structure are open to peer review.

---

## System requirements

| Item | Requirement |
|------|-------------|
| OS | Windows 10 / 11 (64-bit) |
| RAM | 16 GB or more recommended (video, sequences, multi-frame SR) |
| Disk | SSD recommended; transcode and stream capture need cache space |
| GPU | CPU/OpenCV pipeline; **no GPU or AI dependency** |
| Other | ffmpeg bundled or on PATH for DVR conversion and RTSP/HTTP recording |

---

## Positioning

Claritas is a **classic-algorithm forensic workstation**, aimed at roughly **80% of common surveillance and law-enforcement workflows** with **Amped-style auditability**, not a feature-for-feature clone of any single international product or a claim of inherited courtroom brand recognition.

**We deliver:**

- Reproducible filter chains + full evidence-chain reports
- Classic, explainable, defensible image and video analysis
- Case-centric workflow, multilingual UI, and structured documentation

**We do not deliver:**

- One-click AI enhancement or generative restoration
- A promise to fully replace a specific competitor SKU

---

## Contact

For **demos, licensing, volume deployment, or training**, contact us:

- **Website:** [your URL]
- **Email:** [your email]
- **Phone:** [your phone]

---

*Claritas Forensic Workstation · Version 1.0.1*
*Clarity you can reproduce. Evidence you can defend.*