Feature amount generation device, feature amount generation method, and non-transitory medium saving program

Patent metadata

Field	Value
Publication number	US-9798950-B2
Application number	US-201514795510-A
Country	US
Kind code	B2
Filing date	Jul 9, 2015
Priority date	Jul 9, 2015
Publication date	Oct 24, 2017
Grant date	Oct 24, 2017

Abstract

A feature amount generation device includes a feature point detecting unit which detects a feature point from an input image, a pixel information generating unit which generates pixel information from pixel values of the surroundings of the feature point in the input image, a feature point pixel information distribution generating unit which generates a first distribution of pixel information of a first region including the feature point, a sub-feature point detecting unit which detects sub-feature points that are feature points in the surroundings of the feature point, a sub-feature point pixel information distribution generating unit which generates second distributions of pixel information of second regions including the sub-feature points, and a feature amount generating unit which generates a feature amount at the feature point based on the first distribution and the second distributions.

First claim

What is claimed is:

1. A feature amount generation device comprising:
a feature point detecting unit configured to detect a feature point from an input image;
a pixel information generating unit configured to generate pixel information from pixel values of surroundings of the feature point in the input image;
a feature point pixel information distribution generating unit configured to generate a first distribution of pixel information of a first region including the feature point;
a sub-feature point detecting unit configured to detect sub-feature points which are feature points in the surroundings of the feature point;
a sub-feature point pixel information distribution generating unit configured to generate second distributions of pixel information of second regions including the sub-feature points; and
a feature amount generating unit configured to generate a feature amount at the feature point based on the first distribution and the second distributions,
wherein the sub-feature point pixel information distribution generating unit sets points of which feature evaluation values are lower than a feature evaluation value of the feature point detected by the feature point detecting unit as candidates for the sub-feature points.

2. The feature amount generation device according to claim 1 ,
wherein the sub-feature point pixel information distribution generating unit determines a number of partitions of the second regions according to a number of the sub-feature points detected by the sub-feature point detecting unit, partitions the second regions so that a number of regions after the partition becomes the determined number of partitions, and generates a second distribution for every region after the partition.

3. The feature amount generation device according to claim 1 ,
wherein the sub-feature point pixel information distribution generating unit determines a partition size of the second regions according to a number of the sub-feature points detected by the sub-feature point detecting unit, partitions the second regions so that a size of regions after the partition becomes the determined partition size, and generates a second distribution for every region after the partition.

4. The feature amount generation device according to claim 1 ,
wherein the sub-feature point detecting unit determines one point or a plurality of points of which feature evaluation values are higher than a predetermined value among the candidates for the sub-feature points as the sub-feature points.

5. The feature amount generation device according to claim 1 ,
wherein the feature amount generating unit sets, as a feature amount at the feature point, a set in which the first distribution and a distribution obtained by arranging the second distributions related to the plurality of sub-feature points in order based on feature evaluation values of the plurality of sub-feature points are combined.

6. The feature amount generation device according to claim 1 ,
wherein the pixel information generating unit generates gradient vectors including angles and lengths of gradients of the pixel values of the surroundings of the feature point in the input image as the pixel information.

7. The feature amount generation device according to claim 6 ,
wherein the sub-feature point pixel information distribution generating unit generates histograms showing distributions of cumulative totals of lengths with respect to angles as the second distributions by regarding gradient vectors of pixels of the second regions among the gradient vectors generated by the pixel information generating unit as elements.

8. The feature amount generation device according to claim 7 ,
wherein the feature point pixel information distribution generating unit generates a histogram showing a distribution of cumulative totals of lengths with respect to angles as the first distribution by regarding gradient vectors of pixels of the first region among the gradient vectors generated by the pixel information generating unit as elements.

9. The feature amount generation device according to claim 1 ,
wherein the pixel information generating unit generates the pixel information using at least one of luminance values and color information of the input image as the pixel values.

10. The feature amount generation device according to claim 1 ,
wherein the pixel information generating unit generates the pixel information using a raw image as the input image.

11. The feature amount generation device according to claim 1 , further comprising
a main direction adjusting unit configured to calculate a main direction of the feature point based on gradient angles of the pixel values of respective pixels in the surroundings of the feature point in the input image, and rotates the input image according to the main direction,
wherein the pixel information generating unit generates pixel information from pixel values of surroundings of the feature point in the input image rotated by the main direction adjusting unit.

12. A feature amount generation method comprising:
a step of detecting a feature point from an input image;
a step of generating pixel information from pixel values of surroundings of the feature point in the input image;
a step of generating a first distribution of pixel information of a first region including the feature point;
a step of detecting sub-feature points which are feature points in the surroundings of the feature point;
a step of generating second distributions of pixel information of second regions including the sub-feature points;
a step of setting points of which feature evaluation values are lower than a feature evaluation value of the feature point detected in the step of detecting the feature point from the input image as candidates for the sub-feature points; and
a step of generating a feature amount at the feature point based on the first distribution and the second distributions.

13. A non-transitory medium saving a program for causing a computer to perform:
a step of detecting a feature point from an input image;
a step of generating pixel information from pixel values of surroundings of the feature point in the input image;
a step of generating a first distribution of pixel information of a first region including the feature point;
a step of detecting sub-feature points which are feature points in the surroundings of the feature point;
a step of generating second distributions of pixel information of second regions including the sub-feature points;
a step of setting points of which feature evaluation values are lower than a feature evaluation value of the feature point detected in the step of detecting the feature point from the input image as candidates for the sub-feature points; and
a step of generating a feature amount at the feature point based on the first distribution and the second distributions.

Assignees

• Olympus Corp

Inventors

• Nakashima Aya
• Kohashi Atsushi
• Sonoda Yasuko

Classifications

• G06V10/50 Primary

  by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis · CPC title

• G06V10/462

  Salient features, e.g. scale invariant feature transforms [SIFT] · CPC title

• G06T3/4069

  by subpixel displacements · CPC title

• G06T3/60

  Rotation of whole images or parts thereof · CPC title

• G06K9/4642

  Physics · mapped topic