Person Identification Under Variable Illumination Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing person identification processing systems struggle to maintain accuracy when the illumination state changes drastically, such as in a stage environment where lighting conditions vary rapidly, leading to difficulties in tracking performers during live concerts.
Innovation Solution
An information processing device equipped with a person identification unit that determines whether the illumination state at the time of image capture is similar to the illumination state during the learning process, and only executes person identification processing when the similarity is confirmed, using feature information acquired under similar illumination conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If person identification processing is executed using feature data acquired in advance under different illumination states, then the system can operate across various lighting conditions, but the collation degree decreases and identification accuracy deteriorates
Solution Approach 1:
The system performs learning processing to acquire feature data in advance under multiple predetermined illumination states (e.g., bright, dim, colored lighting). This preliminary action enables the system to have identification models ready for various lighting conditions before actual person identification is needed, resolving the contradiction by preparing adaptability in advance while maintaining accuracy through condition-specific feature data.
Solution Approach 2:
The system changes the parameter of illumination state by acquiring and storing feature data under different lighting conditions (brightness, color temperature, direction). This parameter variation approach allows the system to select the most appropriate feature data for the current illumination state, thereby maintaining high identification accuracy across diverse lighting environments while adapting to different conditions.
2Area of stationary object
If the camera tracks a performer who moves around on the stage, then comprehensive coverage is achieved, but the continuous tracking becomes difficult and the camera loses sight of the target
Solution Approach 1:
The system extracts and stores feature data of the tracking target at multiple positions and illumination states encountered during movement. When the camera loses sight of the target, the system retrieves the previously extracted feature data corresponding to the current illumination state to resume tracking, thus maintaining reliability while allowing broad coverage area.
Solution Approach 2:
The system performs preliminary feature data acquisition at various positions and lighting conditions before tracking interruption occurs. This advance preparation of position-specific and illumination-specific feature data enables quick recovery and continuous tracking even when the target moves out of view, resolving the contradiction between coverage area and tracking reliability.
3Loss of time
If learning processing is performed once at the start of a concert, then processing time is reduced, but the illumination state during subsequent live concert periods differs from learning conditions, reducing collation degree
Solution Approach 1:
The system performs learning processing in advance under multiple predetermined illumination states that match expected concert lighting conditions. By preparing feature data for various lighting scenarios beforehand, the system eliminates the need for real-time learning during the concert, thus minimizing processing time loss while maintaining high collation degree through condition-matched feature data selection.
Solution Approach 2:
The system varies the illumination state parameter during preliminary learning processing to cover the range of conditions expected during the concert. This approach allows the system to maintain high collation degree across different lighting conditions while keeping processing time minimal, as the system simply selects from pre-acquired feature data rather than performing new learning during the event.
Data Source
AI summary
To enable highly accurate person identification processing by executing person identification processing in a case where it is confirmed that an illumination state at the time of capturing an image is similar to an illumination state at the time of capturing a learning image from which collation feature information has been acquired. A person identification unit that inputs a camera-captured image, acquires feature information for person identification from the input image, and executes collation processing between the acquired feature information and the collation feature information to execute the person identification processing is included. The person identification unit determines whether or not the illumination state at the time of capturing the input image is similar to the illumination state at the time of capturing the learning image from which the collation feature information has been acquired with reference to an illumination control program generated in advance, and executes the person identification processing to which the input image is applied in a case where the similarity is determined.