Checkout Reading Device Using ToF Depth for Basket Item Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sales registration processing in retail transactions is inefficient as it requires operators to individually photograph and identify each commodity using generic object recognition, which can be improved by capturing an entire basket of commodities at once.
Innovation Solution
A reading device equipped with a time-of-flight camera captures images of a basket from above, identifies the uppermost commodity using distance information and generic object recognition, and adjusts for rearrangements to ensure efficient registration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If operators photograph each commodity individually using generic object recognition, then object identification accuracy is maintained, but sales registration processing time increases
Solution Approach 1:
The patent segments the sales registration process into two phases: bulk capture of all commodities in the basket using a snapshot image, followed by individual identification of each commodity through generic object recognition. This segmentation allows simultaneous processing of multiple objects while maintaining individual identification accuracy, thereby reducing overall processing time without sacrificing precision.
2Reliability
If operators hold each commodity over the image capturing area for individual imaging, then clear object detection is achieved, but operational complexity and time consumption increase
Solution Approach 1:
The patent extracts the image capturing function from the manual handling process. Instead of requiring operators to manually position each commodity, the system captures a snapshot image of the entire basket contents, automatically extracts individual commodity images from the snapshot, and performs recognition. This extraction of the capturing function from manual operation simplifies the operator's task while maintaining detection reliability through automated image processing.
3Productivity
If the reading device captures the entire basket at once, then processing efficiency improves, but difficulty in identifying individual commodities increases
Solution Approach 1:
The patent performs preliminary action by capturing a snapshot image of the entire basket before individual identification is needed. This snapshot serves as a preliminary record containing all commodity information, which is then processed to extract individual commodity images for recognition. This preliminary capture enables efficient bulk processing while maintaining the ability to identify individual commodities through subsequent automated extraction and recognition processes.
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
Enables efficient sales registration by automatically identifying and registering commodities in a basket without manual repositioning, enhancing processing efficiency and accuracy.
Implementation Method 1
A reading device equipped with a time-of-flight camera captures images of a basket from above, identifies the uppermost commodity using distance information
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A reading device includes display and an image capturing device to image objects on a checkout counter. A control unit receives a first image including at least one object on the checkout counter, extracts a first object that is closest to the image capturing device in the first image, then displays an image of the first object and a frame around the first object. Generic object recognition is performed on the first object. A second image is received from the image capturing unit. The control unit determines whether the first object is still closest to the image capturing device. A second object is extracted from the second image if the first object is no longer closest to the image capturing device and the second object is the one closest to the image capturing device in the second image.