Stereo Image Parallax Matching With Concatenated Pixel Memory
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods for distance measurement, such as template matching, face challenges with large circuit scales due to pixel data storage requirements and increased data access leading to decreased throughput.
Innovation Solution
An image processing apparatus that includes an image acquisition unit, correlation acquisition unit, parallax amount calculation unit, and a storage unit that concatenates and stores pixel data in a memory to reduce the need for large register circuits and optimize data access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If pixel data is secured on a register circuit for template matching, then template matching can be performed, but the circuit scale becomes large
Solution Approach 1:
The patent divides the pixel data storage function into two parts: a small register circuit that holds only the fiducial region data needed for template matching, and a main memory that stores the entire image data. This segmentation allows template matching to be performed while keeping the register circuit scale small.
Solution Approach 2:
The patent introduces a control unit as an intermediary that manages data transfer between the main memory and the register circuit. The control unit retrieves reference region data from main memory and supplies it to the register circuit when needed for template matching, eliminating the need for the register circuit to store all pixel data permanently.
2Device complexity
If a dedicated memory unit such as SRAM is used to secure several lines of pixel data, then pixel data can be stored externally, but the number of accesses for reading out necessary data increases and throughput decreases
Solution Approach 1:
The patent pre-loads the fiducial region data from the first image into the register circuit before template matching begins. By preparing this data in advance, the system minimizes the number of memory accesses required during the actual template matching process, thereby maintaining high throughput.
Solution Approach 2:
The patent optimizes the register circuit to store only the specific fiducial region data that is locally required for template matching, rather than storing entire lines or frames of pixel data. This localized storage approach reduces memory access requirements while maintaining processing speed.
Data Source
AI summary
An image processing apparatus comprises: an image acquisition unit configured to acquire a first image and a second image having parallax; a correlation acquisition unit configured to acquire a plurality of correlation information between an image of a fiducial region in the first image and images of a plurality of reference regions corresponding to the fiducial region in the second image; a parallax amount calculation unit configured to calculate a parallax amount between the image of the fiducial region and the image of the reference region based on the plurality of correlation information; and a storage unit configured to hold, in memory for a predetermined period, pixel data configuring the first image and the second image. The correlation acquisition unit is further configured to sequentially acquire the pixel data configuring the first image and the second image acquired by the image acquisition unit, concatenate the pixel data with a plurality of past pixel data each time the pixel data is acquired, and store the concatenated pixel data at each address of the memory.