Single Camera Distance Calculation Using Motion Parallax
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The stereo vision method for calculating object distance using disparity between images faces challenges in accurately determining distances due to small pixel differences, especially between distant and near objects, and requires complex image processing for pixel matching, which is cumbersome and inaccurate with limited inter-camera distance.
Innovation Solution
An image distance calculator that extracts frame images from moving images, generates slice images based on temporal pixel changes, uses dynamic programming for pixel matching, and applies the mean-shift method for region segmentation to calculate object distance using a ratio of pixel densities in corresponding regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the inter-camera distance d is increased to increase pixel difference and improve distance calculation accuracy, then measurement precision improves, but it becomes difficult to obtain simultaneous images of the same object from two cameras
Solution Approach 1:
Instead of using two cameras to capture simultaneous images (stereo vision), the patent inverts the approach by using a single camera to capture sequential images at different positions. The camera moves along the optical axis to capture the same object at different distances, transforming the problem from simultaneous dual-camera capture to sequential single-camera capture.
Solution Approach 2:
The patent transitions from the stereo vision approach (using horizontal separation between two cameras) to motion parallax approach (using temporal separation with camera movement along the optical axis). This adds the time dimension to the image capture process, allowing distance calculation through sequential imaging rather than simultaneous imaging.
2Measurement precision
If the inter-camera distance d is increased to increase pixel difference, then measurement precision improves, but device complexity increases due to difficulty in matching corresponding pixels
Solution Approach 1:
The patent performs preliminary actions by capturing multiple sequential images at different camera positions before performing distance calculation. This allows the system to accumulate sufficient pixel displacement information across multiple frames, making the subsequent matching and calculation processes more straightforward and accurate.
Solution Approach 2:
The patent utilizes continuous camera movement along the optical axis to capture a series of images showing progressive displacement of object pixels. This continuous action ensures that corresponding pixels can be tracked across multiple frames, providing a continuous basis for accurate distance calculation without requiring complex simultaneous processing.
3Ease of operation
If a single camera is used to capture moving images for distance calculation, then device complexity is reduced and ease of operation improves, but measurement precision may be compromised compared to stereo vision methods
Solution Approach 1:
The patent employs periodic action by capturing images at regular intervals as the camera moves along the optical axis. This periodic sampling ensures that sufficient displacement information is collected at consistent time intervals, enabling accurate distance calculation while maintaining simple single-camera operation.
Solution Approach 2:
The patent utilizes the dynamic movement of the camera along the optical axis to create measurable pixel displacement in sequential images. By making the camera position dynamic rather than static, the system generates the necessary parallax effect for distance calculation using a single camera, achieving both operational simplicity and measurement accuracy.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach simplifies image capture by allowing distance calculation from a single camera, reducing the burden of capturing images with multiple cameras and achieving higher accuracy in distance measurement across a wider dynamic range compared to traditional methods.
Implementation Method 1
a distance calculating section that calculates a distance from the camera to the object on the basis of the pixel displacement
Data Source
AI summary
In an image distance calculator (100), a CPU (104) extracts a frame image from moving images of an object captured by a camera, generates a slice image on the basis of a temporal change in a pixel line on a y-axis at a point x0 in the frame image, calculates a spotting point on the basis of correspondences between pixels in the slice image and pixels in the frame image, obtains pixels in the frame image corresponding to pixels in the slice image by a back-trace process, segments the frame image and slice image into regions, determines a corresponding region corresponding to a segmented region of the slice image, calculates a ratio value from an average q of the numbers of pixels in the corresponding region in the frame image and an average p of the numbers of pixels in the segmented region of the slice image, and calculates the distance z from the camera to the object for each corresponding region using a predetermined distance function.