Crate oximedia_cv 

Computer vision module for OxiMedia.

oximedia-cv provides computer vision algorithms and image processing capabilities for the OxiMedia multimedia framework. This includes:

• Image processing: Resize, color conversion, filtering, edge detection
• Detection: Face detection, motion detection, object detection, corner detection
• Transforms: Affine and perspective transformations
• Enhancement: Super-resolution and denoising using AI models
• Machine Learning: ONNX Runtime integration for ML model inference
• Tracking: Optical flow, feature tracking, object tracking
• Stabilization: Video stabilization with motion smoothing
• Scene detection: Video scene changes and shot boundary detection
• Quality: Video quality assessment metrics (PSNR, SSIM, VMAF)
• Interpolation: Frame interpolation using optical flow for frame rate conversion
• Chroma key: Green screen and blue screen processing with spill suppression
• Content-aware scaling: Seam carving and intelligent resizing
• Interlace detection: Interlacing and telecine detection with IVTC recommendations
• Motion blur: Motion blur synthesis and removal with deconvolution algorithms
• Fingerprinting: Perceptual content fingerprinting for video and audio

Modules

• image: Image processing operations (resize, filter, histogram, etc.)
• detect: Detection algorithms (face, motion, object, corner)
• transform: Geometric transformations (affine, perspective)
• enhance: Image enhancement (super-resolution, denoising)
• ml: Machine learning and ONNX Runtime integration
• tracking: Video tracking and optical flow (incl. enhanced SORT with tracking::SortTrackerV2 / tracking::LkTracker)
• stabilize: Video stabilization with motion smoothing
• scene: Video scene detection and shot boundary detection
• quality: Video quality metrics (PSNR, SSIM, VMAF, temporal)
• interpolate: Frame interpolation using optical flow for smooth motion
• chroma_key: Chroma keying (green screen) with spill suppression and compositing
• scale: Content-aware scaling using seam carving and saliency detection
• interlace: Interlacing and telecine detection for video content analysis
• motion_blur: Motion blur synthesis and removal with deconvolution
• fingerprint: Perceptual fingerprinting for content identification and matching

Example

use oximedia_cv::image::{ResizeMethod, ColorSpace};
use oximedia_cv::detect::BoundingBox;
use oximedia_cv::transform::AffineTransform;

// Example usage
let bbox = BoundingBox::new(10.0, 20.0, 100.0, 150.0);
assert!(bbox.area() > 0.0);

Re-exports

pub use error::CvError;

pub use error::CvResult;

Modules

action_recognition

Action recognition using temporal feature analysis.

background_subtraction

Background subtraction via a per-pixel Mixture-of-Gaussians (MOG2-inspired) model.

blob_detector

Blob detection via connected-component labeling and morphological filtering.

bounding_box

Axis-aligned bounding-box utilities for computer-vision pipelines.

chroma_key

Chroma key (green screen) processing module.

color_cluster

Color clustering via K-Means for dominant color extraction.

contour

Contour finding and analysis for binary/edge images.

deep_sort

DeepSORT multi-object tracker combining Kalman filter motion prediction with deep appearance feature embeddings for robust re-identification.

depth_estimation

Stereo depth estimation from disparity maps.

detect

Detection algorithms.

enhance

Image enhancement module.

error

Error types for computer vision operations.

feature_extract

Feature extraction: SIFT-like keypoints, ORB-like descriptors, and BRIEF.

feature_match

Feature descriptor matching utilities for stereo vision, SLAM, and object tracking.

fingerprint

Content fingerprinting for video and audio.

histogram_backproject

Histogram back-projection for object localization in images.

hough_transform

Hough transform for detecting lines and circles in images.

image

Image processing operations.

instance_segmentation

Instance segmentation: connected-component labeling with per-object binary masks.

interlace

Interlacing and telecine detection for video content.

interpolate

Video frame interpolation using optical flow.

keypoint

Keypoint detection and matching for feature-based computer vision tasks.

lane_detect

Lane detection for video frames.

lens_distortion

Lens distortion correction (barrel, pincushion, fish-eye).

morphology

Mathematical morphology operations for binary and grayscale images.

motion_blur

Motion blur synthesis and removal.

motion_vector

Motion vector estimation between video frames.

obj_tracking

Object tracking: centroid tracking, IoU-based track association, and track lifecycle management.

optical_flow_field

Dense optical flow field representation and analysis.

panorama_stitch

Simple offset-based panorama stitching with linear alpha blending.

pose_estimation

Human pose estimation module.

quality

Video quality assessment metrics.

registration

Video alignment and registration tools.

scale

Content-aware video scaling for OxiMedia.

scene

Video scene detection and shot boundary detection.

segmentation

Image segmentation: connected components, region growing, and a watershed-like approximation.

stabilize

Video stabilization with motion smoothing.

style_transfer

CPU-based artistic style transfer using neural-style approximation.

superpixel

Superpixel segmentation using a simplified SLIC algorithm.

text_detect

Text detection and recognition (OCR) for extracting burned-in subtitles.

texture_analysis

Image texture analysis using Gray-Level Co-occurrence Matrices (GLCM).

tracking

Video tracking and motion estimation.

transform

Geometric transformation module.

video_matting

Video matting — alpha matte extraction without a chroma key.