Image Texture Classification Using Fused Bitmaps for Object Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems for image classification in computing systems struggle to accurately distinguish between textured and textureless images, which is crucial for automated object recognition and robot interaction, especially in environments with varying lighting conditions and object surfaces.
Innovation Solution
A computing system that generates bitmaps from image portions to identify visual features and intensity variations, using a fused bitmap to classify images as textured or textureless, and performs motion planning for robot interaction based on these classifications, while compensating for lighting effects like glare and shadows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image classification methods are used, then the system can process images quickly, but the accuracy of distinguishing textured from textureless images deteriorates under varying lighting conditions
Solution Approach 1:
The patent segments the image processing into multiple specialized bitmaps: visual feature bitmap (detecting edges, corners, textures), intensity variation bitmap (detecting lighting gradients), and fusion bitmap (combining both). This segmentation allows each bitmap to specialize in detecting specific features while being less sensitive to lighting variations, thereby improving classification accuracy under varying lighting conditions.
Solution Approach 2:
The patent transforms the image data into different parameter representations through bitmap generation. By converting visual features and intensity variations into binary bitmap formats, the system changes the parameter space from continuous pixel values to discrete feature presence/absence indicators, making the classification more robust to lighting intensity changes while maintaining detection accuracy.
2Measurement precision
If multiple bitmaps are generated for feature detection, then the classification accuracy improves, but the computational complexity increases
Solution Approach 1:
The patent generates multiple bitmaps (visual feature bitmap, intensity variation bitmap, and fusion bitmap) to perform partial analysis of different image characteristics. Rather than using a single complex classifier, the system performs multiple simpler binary transformations and combines them, achieving high classification accuracy through cumulative partial actions that are computationally efficient.
Solution Approach 2:
The patent merges the results of visual feature detection and intensity variation detection into a fusion bitmap. This combining step integrates multiple feature sources to improve classification accuracy while maintaining computational efficiency by using simple logical operations to merge the binary bitmap results rather than complex integration algorithms.
3Measurement precision
If the system classifies all image portions in detail, then object recognition accuracy improves, but the processing time increases
Solution Approach 1:
The patent divides the image into multiple portions and processes each portion independently through the bitmap generation pipeline. This segmentation allows parallel processing of different image regions, maintaining high recognition accuracy for each object while reducing overall processing time through concurrent execution of bitmap operations on multiple image segments.
Data Source
AI summary
Systems and methods for classifying at least a portion of an image as being textured or textureless are presented. The system receives an image generated by an image capture device, wherein the image represents one or more objects in a field of view of the image capture device. The system generates one or more bitmaps based on at least one image portion of the image. The one or more bitmaps describe whether one or more features for feature detection are present in the at least one image portion, or describe whether one or more visual features for feature detection are present in the at least one image portion, or describe whether there is variation in intensity across the at least one image portion. The system determines whether to classify the at least one image portion as textured or textureless based on the one or more bitmaps.