Abstract
This work introduces a novel local patch descriptor that remains invariant under varying conditions of orientation, viewpoint, scale, and illumination. The proposed descriptor incorporate polynomials of various degrees to approximate the local patch within the image. Before feature detection and approximation, the image micro-texture is eliminated through a guided image filter with the potential to preserve the edges of the objects. The rotation invariance is achieved by aligning the local patch around the Harris corner through the dominant orientation shift algorithm. Weighted threshold histogram equalization (WTHE) is employed to make the descriptor in-sensitive to illumination changes. The correlation coefficient is used instead of Euclidean distance to improve the matching accuracy. The proposed descriptor has been extensively evaluated on the Oxford’s affine covariant regions dataset, and absolute and transition tilt dataset. The experimental results show that our proposed descriptor can categorize the feature with more distinctiveness in comparison to state-of-the-art descriptors.
Highlights
Feature descriptor is used to describe the image interest region in such a way that the description remains robust against geometric and photometric transformations of the image
To bring improvement in the local patch representation under varying conditions of illumination and 3D view angle variation, image blur, and JPEG compression noise in the picture, we have introduced Polynomial Approximation of Local Surface (PALS) which provide a more distinctive representation of the local features
If we reduce the amount of the threshold, we get many orientations for the patch alignment, which is not a realistic approach
Summary
Feature descriptor is used to describe the image interest region in such a way that the description remains robust against geometric and photometric transformations of the image. Histogram of oriented gradient based descriptors are scale and rotation invariant [12] They are computationally expensive [13], in-case of illumination difference and image blur, their performance degrades. This work focuses on the development of an improved image local descriptor, which is robust to compression noise, blurriness, illumination, and view-angle variations. The maximally stable extremal regions (MSER) [34] is a rotation, scale, and affine invariant, which is used to detect the feature points with high repeatability and efficiency, but it is sensitive to image blur. To bring improvement in the local patch representation under varying conditions of illumination and 3D view angle variation, image blur, and JPEG compression noise in the picture, we have introduced Polynomial Approximation of Local Surface (PALS) which provide a more distinctive representation of the local features. The feature vectors matching technique is modified to improve the classification accuracy
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
