Fisheye Camera Object Detection via Image Rectification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object detection systems are not invariant to radial distortion present in fisheye images, making them unsuitable for use with fisheye cameras commonly used in vehicles, which are essential for improving safety by detecting pedestrians and other objects during reversing maneuvers.
Innovation Solution
An object detection system integrated with fisheye lens cameras that classifies objects of interest, estimates their distance, and applies bounding boxes around them, using camera calibration information and a processor to transmit this data for display and potential vehicle control, including sound-based warnings.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional object detection systems are used, then they can detect objects in standard images, but they cannot accurately detect objects in fisheye images with radial distortion
Solution Approach 1:
The patent transforms the fisheye image parameters by applying a rectification process that converts radial distortion to a rectilinear projection. This changes the geometric parameters of the image to match standard camera models, allowing conventional object detection algorithms to function accurately on fisheye captured data.
Solution Approach 2:
The patent introduces an intermediary rectification step between image capture and object detection. This intermediary process converts the fisheye image into a rectilinear representation that serves as a bridge, enabling standard detection algorithms to work with fisheye camera inputs without modification.
2Area of stationary object
If fisheye lens cameras are used to capture wide field of view, then coverage area increases, but image distortion increases making conventional detection inapplicable
Solution Approach 1:
The rectification process changes the geometric parameters of the fisheye image by applying a transformation that maps the circular fisheye projection to a rectangular rectilinear projection. This parameter transformation preserves the wide field of view coverage while eliminating the radial distortion that complicates detection.
Solution Approach 2:
The patent segments the fisheye image processing into distinct stages: capture with wide FOV, rectification transformation, and conventional detection. This segmentation allows each stage to optimize for its specific function - the camera captures wide area, the rectification corrects distortion, and standard detectors perform object recognition.
3Measurement precision
If image rectification is applied to remove radial distortion, then object detection accuracy improves, but processing complexity increases
Solution Approach 1:
The rectification process applies a deterministic parameter transformation using pre-computed lookup tables or analytical formulas. While this adds a processing step, the transformation itself is computationally efficient and can be pre-calibrated, making the complexity manageable compared to the gain in measurement precision.
Data Source
AI summary
An object detection system for used in a vehicle includes an object detector is provided. One or more fisheye lens cameras coupled to an object detection system are positioned at various location of the vehicle for capturing a field of view (FOV) into an image. The image is split into multiple different set of perspective images. Each perspective images may include a portion of overlap having common or identical object of interest. The object detector classifies various objects of interest in the perspective images, estimates distance of objects of interest from the vehicle using camera calibration information stored in one of the camera, and transmits the sensed information that corresponds to the distance of interest to a processor. The processor includes sequence of instruction or object code either located in one of the object detection system, in-vehicle network, and an electronic device processes the sensed information and applies bounding boxes around objects of interest the vehicle or in the electronic device. One or more images incorporated the colored bounding boxes are transmitted to a display unit or screen for display as human readable format.