Multi-Camera Object Recognition Under Occlusion and Glare
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile robots equipped with cameras face difficulties in identifying inventory when products are overlapped or obscured by obstacles, and issues with bright lighting conditions lead to deteriorated image recognition rates due to excessive light reflection.
Innovation Solution
An electronic apparatus with multiple cameras having different image capturing angles, where a processor acquires and combines images to identify obscured portions of a target object, and adjusts light emission based on identified regions of interest to enhance recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single camera is used to capture product images, then the device complexity is low, but the ability to identify obscured target objects deteriorates
Solution Approach 1:
The patent transitions from a single-camera system to a multi-camera system with different image capturing angles. Specifically, it uses a first camera with a first image capturing angle and a second camera with a second image capturing angle greater than the first, enabling identification of target objects obscured in one view but visible in another dimension/angle.
Solution Approach 2:
The patent divides the imaging task into multiple segments by using multiple cameras with different fields of view. The first camera captures images at a narrower angle while the second camera captures images at a wider angle, allowing the system to segment the observation space and combine results for complete target identification.
2Illumination intensity
If a flash is used to illuminate dark indoor spaces, then the illumination intensity increases, but light reflection from products and labels deteriorates image recognition
Solution Approach 1:
The patent applies local quality by using a light emitter that provides localized illumination rather than omnidirectional flash lighting. The light emitter illuminates specific regions of interest (such as shelves or product areas) with controlled intensity, reducing excessive reflection from product surfaces and labels while maintaining sufficient brightness for recognition.
3Measurement precision
If multiple cameras with different image capturing angles are used, then the ability to identify obscured target objects improves, but the device complexity increases
Solution Approach 1:
The patent implements universality by designing a mobile robot platform that integrates multiple cameras with different image capturing angles into a single multifunctional system. The first camera (narrower angle) and second camera (wider angle) work together to provide both detailed and contextual views, enabling the system to handle various identification scenarios with a unified device.
Data Source
AI summary
An electronic apparatus is disclosed. The electronic apparatus comprises: a first camera having a first image capturing angle; a second camera having a second image capturing angle larger than the first image capturing angle; and a processor which when a portion of a target object is identified and an obstacle object overlapping the other portion of the target object is identified from at least one of a first captured image acquired from the first camera and a second captured image acquired from the second camera, acquires location information of the identified obstacle object, identifies a region of interest in the second captured image on the basis of the acquired location information of the obstacle object, and identifies the other portion of the target object on the basis of the region of interest in the second captured image.