Image Adaptive Feature Extraction Method and Application Thereof

Patent metadata

Field	Value
Publication number	US-2020160088-A1
Application number	US-201916676453-A
Country	US
Kind code	A1
Filing date	Nov 7, 2019
Priority date	Nov 19, 2018
Publication date	May 21, 2020
Grant date	—

Abstract

An image adaptive feature extraction method includes dividing an image into a plurality of blocks, performing a feature extraction processing on the plurality of blocks, and obtaining a block feature from each of the plurality of blocks after the feature extraction processing; calculating each block feature by means of a support vector machine (SVM) classifier, wherein each block feature is calculated to obtain a hyperplane normal vector; setting a threshold value, determining the block feature according to the hyperplane normal vector, recording the block as an adaptive feature block when a value of the hyperplane normal vector is higher than the threshold value, and integrating each adaptive feature block to form an adaptive feature image. Because an image adaptive feature extraction process is performed before a pedestrian image detection is calculated, and effective feature data is then selected, computational efficiency is boosted and detection pedestrian error probability is reduced.

First claim

What is claimed is:

1 . An image adaptive feature extraction method, comprising steps of:
(A) dividing an image into a plurality of blocks, performing a feature extraction processing on the plurality of blocks, and obtaining a block feature from each of the plurality of blocks after the feature extraction processing; (B) calculating each block feature by means of a support vector machine (SVM) classifier, wherein each block feature is calculated to obtain a hyperplane normal vector; and (C) setting a threshold value, determining the block feature according to the hyperplane normal vector, recording the block as an adaptive feature block when a value of the hyperplane normal vector is higher than the threshold value, and integrating each adaptive feature block to form an adaptive feature image.

2 . The image adaptive feature extraction method according to claim 1 , wherein the feature extraction processing is a histogram of oriented gradients (HOG), a local binary pattern (LBP), or a histogram of local intensity difference (HLID).

3 . A pedestrian thermal image detection method, comprising steps of:
(A) reading a raw thermal image, wherein the raw thermal image comprises a specific ambient information; (B) dividing the raw thermal image into a plurality of blocks, performing a pedestrian feature extraction processing on the plurality of blocks, and obtaining a block feature from each of the plurality of blocks after the pedestrian feature extraction processing; (C) calculating each block feature by means of a support vector machine (SVM) classifier, wherein each block feature is calculated to obtain a hyperplane normal vector; (D) setting a threshold value, determining the block feature according to the hyperplane normal vector, recording the block as an adaptive feature block when a value of the hyperplane normal vector is higher than the threshold value, and integrating each adaptive feature block to form an pedestrian feature image; and (E) performing a pedestrian image detection by means of the pedestrian feature image.

4 . The pedestrian thermal image detection method according to claim 3 , wherein the pedestrian image detection is a histogram of oriented gradients (HOG).

5 . The pedestrian thermal image detection method according to claim 3 , wherein the pedestrian image detection is a local binary pattern (LBP).

6 . The pedestrian thermal image detection method according to claim 3 , wherein the pedestrian image detection is a histogram of local intensity difference (HLID).

7 . The pedestrian thermal image detection method according to claim 3 , wherein the specific ambient information comprises a pedestrian image.

8 . The pedestrian thermal image detection method according to claim 3 , wherein the support vector machine (SVM) classifier is trained by using static humanoid sample data as a training sample database.

9 . The pedestrian thermal image detection method according to claim 3 , wherein the support vector machine (SVM) classifier is trained by using a probe-six dataset as a test database.

Assignees

• Nat Chung Shan Inst Science & Tech

Inventors

• Huang Shih-Shinh
• Chien Shih-Che
• Chang Feng-Chia
• Hsiao Yu-Sung
• Hsiao Chien-Hao

Classifications

• G06T7/223

  using block-matching · CPC title

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title

• G06K9/4642 Primary

  Physics · mapped topic

• G06K9/4671

  Physics · mapped topic

• G06T7/73 Primary

  using feature-based methods · CPC title