Image feature amount output device, image recognition device, the image feature amount output program, and image recognition program

Patent metadata

Field	Value
Publication number	US-11256950-B2
Application number	US-201816480960-A
Country	US
Kind code	B2
Filing date	Jan 31, 2018
Priority date	Jan 31, 2017
Publication date	Feb 22, 2022
Grant date	Feb 22, 2022

Abstract

An image processing device converts an image that is a recognition object image to high-resolution, medium-resolution, and low-resolution images. The device sets the pixel of interest of the high-resolution image, and votes the co-occurrence in a gradient direction with offset pixels, the co-occurrence in the gradient direction pixels in the medium-resolution image, and the co-occurrence in the gradient direction pixels in the low-resolution image, to a co-occurrence matrix. The device creates such a co-occurrence matrix for each pixel combination and for each resolution. The device executes the process on each of the pixels of the high-resolution image, and creates a co-occurrence histogram wherein the elements of a plurality of co-occurrence matrices are arranged in a line. The device normalizes the co-occurrence histogram and extracts, as a feature quantity of the image, a vector quantity having as a component a frequency resulting from the normalization.

First claim

The invention claimed is:

1. An image feature amount output device comprising a processor programmed to:
receive an input of an image;
convert a resolution of the input image;
acquire luminance gradient directions of pixels included in the input image and the image obtained by converting the resolution;
acquire a combination of two pixels from the pixels included in the input image and the image obtained by converting the resolution;
acquire a co-occurrence in the acquired luminance gradient direction with respect to the two pixels related to the acquired combination, using the acquired luminance gradient direction of each pixel;
acquire distribution of an occurrence frequency of the acquired co-occurrence and change the acquired combination of the two pixels; and
output a feature amount of the image as the acquired distribution of the occurrence frequency of the acquired co-occurrence, which does not include an output of the acquired luminance gradient direction.

2. The image feature amount output device according to claim 1 , wherein the processor is programmed to acquire at least a combination of adjacent pixels over the whole image.

3. The image feature amount output device according to claim 1 , wherein the processor is programmed to acquire a combination of pixels having different resolutions.

4. The image feature amount output device according to claim 2 , wherein the processor is programmed to acquire the combination of pixels having different resolutions.

5. The image feature amount output device according to claim 1 , wherein the processor is programmed to acquire a combination of pixels having a same resolution for each resolution.

6. The image feature amount output device according to claim 2 , wherein the processor is programmed to acquire the combination of pixels having a same resolution for each resolution.

7. An image recognition device comprising a processor programmed to:
acquire a reference feature amount expressing a feature amount of a recognizing object with distribution of an occurrence frequency of co-occurrence in a luminance gradient direction;
acquire an object image which is an object to be determined;
input the acquired object image into the image feature amount output device according to claim 1 , and acquire the feature amount of the aforementioned object image;
determine whether the object image includes a recognizing object image by comparing the acquired reference feature amount and the acquired feature amount of the object image; and
output a result of the determination.

8. A non-transitory computer-readable storage medium storing an image feature amount output program that, when executed by a processor, performs functions comprising:
an image input function inputting an image;
a resolution conversion function converting a resolution of the input image;
a gradient direction acquiring function acquiring a luminance gradient directions of pixels included in the input image and the image obtained by converting the resolution;
a pixel combination acquiring function acquiring a combination of two pixels from the pixels included in the input image and the image obtained by converting the resolution;
a co-occurrence acquiring function acquiring a co-occurrence in the luminance gradient direction acquired by the gradient direction acquiring function with respect to the two pixels related to the acquired combination, using the acquired luminance gradient direction of each pixel;
an occurrence frequency acquiring function acquiring distribution of an occurrence frequency of the co-occurrence acquired by the co-occurrence acquiring function and changing the combination of the pixel to be acquired in the pixel combination acquiring function; and
an output function outputting a feature amount of the image as the distribution of the occurrence frequency of the co-occurrence acquired by the occurrence frequency acquiring function, which does not include outputting of the acquired luminance gradient direction.

9. A non-transitory computer-readable storage medium storing an image recognition program that, when executed by the processor, performs functions comprising:
a reference feature amount acquiring function acquiring a reference feature amount expressing a feature amount of a recognizing object with distribution of an occurrence frequency of co-occurrence in a luminance gradient direction;
an object image acquiring function acquiring an object image which is an object to be determined;
an object image feature amount acquiring function inputting the acquired object image into the image input means of the image feature amount output device according to claim 1 , and acquiring a feature amount of the object image;
a determination function determining whether the object image includes a recognizing object image by comparing the acquired reference feature amount and the acquired feature amount of the object image; and
a result outputting function outputting a result of the determination.

10. A non-transitory computer-readable storage medium storing an image recognition program that, when executed by the processor, performs functions comprising:
a reference feature amount acquiring function acquiring a reference feature amount expressing a feature amount of a recognizing object with distribution of an occurrence frequency of co-occurrence in a luminance gradient direction;
an object image acquiring function acquiring an object image which is an object to be determined;
an object image feature amount acquiring function inputting the acquired object image into the image input function of the image feature amount output program according to claim 8 , and acquiring a feature amount of the object image;
a determination function for determining whether the object image includes a recognizing object image by comparing the acquired reference feature amount and the acquired feature amount of the object image; and
a result outputting function outputting a result of the determination.

Assignees

• Aisin Corp
• Kyushu Inst Tech

Inventors

• Yamada Hideo
• Kuno Kazuhiro
• Shibata Masatoshi
• Enokida Shuichi
• TAMUKOH Hakaru

Classifications

• G06V20/58 Primary

  Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads · CPC title

• G06V10/56

  relating to colour · CPC title

• G06V10/758

  Involving statistics of pixels or of feature values, e.g. histogram matching · CPC title

• G06T3/40

  Scaling of whole images or parts thereof, e.g. expanding or contracting · CPC title

• G06T7/00

  Image analysis · CPC title