Input Device Object Detection via Peak Labeling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing object detection methods, such as template matching and binarization, are inadequate for recognizing objects like fingers due to variations in thickness, length, and tilt, and are sensitive to threshold values, leading to poor detection accuracy, especially when multiple objects with different heights and shapes are present.

Innovation Solution

An input device with a sensor system that detects approaching objects and specifies adjacent regions based on peak detection positions and label assignment, using a peak position specification part to identify maximum peak detection positions and a label assigning part to assign labels to surrounding detection positions, improving object separation and recognition accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If template matching is used to recognize objects, then objects with fixed patterns can be identified, but objects with varying characteristics (thickness, length, tilt) cannot be accurately recognized

Engineering Contradiction:
Improveobject recognition accuracyVSAvoidadaptability to object variations
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent segments the detection area into multiple detection positions arranged in rows and columns, and further segments objects into multiple regions by specifying adjacent regions for each detected object. This segmentation allows independent analysis of different parts of objects with varying characteristics, enabling accurate recognition without requiring fixed templates.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by using peak detection at specific detection positions and assigning labels to adjacent regions based on local detection data. Each detection position evaluates its own peak condition independently, and objects are recognized based on local characteristics rather than global template matching, accommodating variations in object thickness, length, and tilt.

Inventive Principle:
Principle #3Local quality

2Measurement precision

If binarization with labeling is used to identify objects, then multiple objects can be separated, but detection accuracy deteriorates when objects have different heights and shapes due to threshold sensitivity

Engineering Contradiction:
Improveobject separation accuracyVSAvoiddetection stability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent changes the detection parameter from fixed threshold binarization to peak condition evaluation. Instead of using a fixed threshold value that requires careful tuning, the system detects peaks based on relative comparison at each detection position, making the detection reliable across objects with different heights and shapes without sacrificing separation accuracy.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If a single threshold is used for binarization, then the recognition process is simple, but multiple objects with different characteristics cannot be accurately detected

Engineering Contradiction:
Improverecognition algorithm complexityVSAvoidmulti-object detection accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent segments both the detection area into grid positions and the objects into adjacent regions, allowing each region to be independently labeled. This segmentation enables accurate multi-object detection without requiring complex global threshold tuning, as each local region can be evaluated independently based on peak conditions.

Inventive Principle:
Principle #1Segmentation

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

The device achieves high-precision recognition of multiple objects with different shapes and distances by accurately specifying adjacent regions and reducing noise, thereby enhancing object separation and recognition accuracy.

Implementation Method 1

a sensor that detects an approach or contact of the object by a change of an electrostatic capacitance has been widely used

Methodology Applied
Scientific EffectElectrostatic capacitance change: Capacitance

Data Source

PatentEP3312704B1Input device, and object detection device and method
Publication Date: 2022.05.11 ALPS ALPINE CO LTD
  • EP3312704B1 patent drawingFigure 1
  • EP3312704B1 patent drawingFigure 2
  • EP3312704B1 patent drawingFigure 3

AI summary

Provided is an input device including a sensor part to detect approaching states of objects at detection positions and an adjacent region specification part to specify adjacent regions of the objects based on detection data from the sensor part. The adjacent region specification part specifies a peak detection position, of which a value of the detection data satisfies a certain peak condition among the detection positions, and conducts a label assigning process that assigns a label applied at the peak detection position to one or more detection positions among surrounding detection positions of the peak detection position. The label is not assigned to the one or more detection positions, and the detection data of the one or more detection positions indicate greater than or equal to a first threshold defined based on the detection data of the peak detection position being specified.