Image processing apparatus and method for encoding an image descriptor based on a gradient histogram
A gradient histogram and image description technology, applied in the field of image descriptors, can solve problems such as ignoring redundant patterns and complexity, and achieve the effect of computational complexity and high efficiency
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
no. 1 example
[0085] In a first embodiment of the invention, sub-descriptors are grouped according to their distance from the center of the descriptor.
[0086] For example, one such grouping is in Figure 9 is shown in , where there are three groups, namely, g containing the sub-descriptor with the largest distance to the descriptor center 0 ={v 0 ,v 3 ,v 12 ,v 15}, g containing the subdescriptor with the second largest distance to the descriptor center 1 ={v1 ,v 2 ,v 4 ,v 7 ,v 8 ,v 11 ,v 13 ,v 14}, and g containing the sub-descriptor with the minimum distance to the descriptor center 2 ={v 5 ,v 6 ,v 9 ,v 10}. Within each group, the subdescriptors are sorted in ascending subdescriptor index order, but this is not restrictive and other orders may be used, such as clockwise starting from the topmost leftmost subdescriptor in the group. Within each group, corresponding elements of the group's sub-descriptors are all assigned the same encoding priority.
[0087] Based on thi...
no. 2 example
[0096] In a second embodiment of the invention, sub-descriptors are grouped firstly according to their distance from the center (first condition) and secondly according to their mutual distance (second condition).
[0097] In this context, the distance between sub-descriptors may again take the form of Euclidean distance or another suitable distance metric, such as Manhattan distance.
[0098] A second condition may be, for example, that the group should not contain sub-descriptors whose distance to each other is below a predetermined threshold. Thus, a predetermined threshold may be set to prevent eg grouping adjacent sub-descriptors.
[0099] The purpose of the second condition is to ensure that subdescriptors of different groups are taken from relatively distant positions of the subdescriptor grid whenever possible, thereby increasing the information content of descriptors with a very low number of features. It should be noted that the second condition may not always be sa...
no. 3 example
[0103] In a third embodiment of the present invention, the subdescriptors are grouped according to their distance from the center and the subdescriptors of each group are sorted in an order defined according to their corresponding coding characteristics (such as their corresponding transformations). coded.
[0104] For example, combining image 3 The transformation deployment mode considers the second embodiment of the present invention Figure 17 , it can be seen that within each group, two sub-descriptors are transformed according to transformation A and two sub-descriptors are transformed according to transformation B. Thus, it is possible to set a common encoding order condition that for each group the transform utilization order must be "AABB", i.e. for each sub-descriptor group such that the transform of the first sub-descriptor in the group is transform A, in the group The transform of the second sub-descriptor is also Transform A, the transform of the third sub-descr...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More