Object Recognition Using Stereo Cameras for Wrapped Article Groups
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition systems struggle with recognizing a group of articles wrapped in transparent wrap as a single transport unit due to limited sensing distance of ultrasonic and radio waves, leading to time-consuming scanning and misidentification when transparent coverings are not present on the upper part of the article group.
Innovation Solution
An object recognition apparatus that utilizes stereo cameras to detect individual article areas and computes frequency distributions of article-to-article information to estimate identical article groups, allowing accurate recognition without ultrasonic or radio wave sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ultrasonic wave or radio wave sensing is used to detect coverings, then the presence/absence of covering can be determined, but the sensing distance is short requiring the sensor to be brought close to articles which makes the scanning process time-consuming
Solution Approach 1:
The patent replaces the mechanical/physical sensing system (ultrasonic wave or radio wave sensors) with an optical imaging system (camera). The camera captures images of the articles, and covering presence is determined by analyzing image data such as transparency, reflection patterns, and article grouping, eliminating the need for close-range physical sensing and time-consuming scanning
Solution Approach 2:
The patent introduces image processing algorithms as an intermediary between the camera and the covering detection function. The algorithms analyze image characteristics (transparency, reflection, spatial arrangement) to infer covering presence, enabling accurate detection without direct contact or close-range sensing
2Measurement precision
If ultrasonic wave or radio wave sensing is used to detect coverings, then covering presence can be identified, but the system cannot recognize article groups bound with transparent wrap on side faces as lump objects since no covering is present at the upper part
Solution Approach 1:
The patent transitions from top-down sensing (ultrasonic/radio waves from above) to multi-dimensional image analysis. The camera captures visual information from multiple angles and the system analyzes spatial relationships, transparency patterns, and article arrangements in three-dimensional space, enabling detection of side-face bindings that top-down sensors cannot detect
Solution Approach 2:
The patent makes the recognition system universal by enabling it to handle multiple article configurations: articles with top coverings, articles with side-face bindings, and transparent wrap configurations. The image-based approach adapts to different covering types and positions by analyzing various image features rather than relying on a single sensing modality
Data Source
AI summary
The present invention provides an object recognition device that can recognize a lump of transport target objects (group of the same articles) as a transport unit on the basis of an arrangement of individual articles in an article group. The object recognition device recognizes the group of the same articles, which is a unit of transport in an environment where a plurality of articles exist, the object recognition device characterized by comprising: an input unit that acquires an image of the plurality of articles; an article detection unit that detects, from the image, article areas where the articles exist; and a group-of-same-articles area estimation unit that acquires inter-article area information, which is information related to an arrangement of the article areas, calculates a frequency distribution of the inter-article area information, and estimates an area of the group of the same articles on the basis of the frequency distribution.