Apparatus and method for storing image on basis of input sound signal

Patent metadata

Field	Value
Publication number	US-10150447-B2
Application number	US-201515124350-A
Country	US
Kind code	B2
Filing date	Mar 4, 2015
Priority date	Mar 13, 2014
Publication date	Dec 11, 2018
Grant date	Dec 11, 2018

Abstract

An image storage device based on an input sound signal comprises: an acceleration sensor unit for outputting an acceleration sensor value according to an impact event; a microphone unit for receiving an input of an impact sound according to the impact event; a camera unit for photographing an external image; and a processing unit for, if an impact event is sensed by the acceleration sensor unit and the impact sound inputted by the microphone unit corresponds to a valid sound which is previously set and stored, storing, in a storage unit, the images photographed by the camera unit in a certain time period both before and after the point in time when the impact event occurred.

First claim

The invention claimed is:

1. An image storage device based on an input sound signal, comprising:
an acceleration sensor unit configured to output an acceleration sensor value in response to receipt of an impact event;
a microphone unit configured to receive an impact sound in response to receipt of the impact event;
a camera unit configured to capture an external image; and
in response to detection of the impact event through the acceleration sensor unit, a processing unit configured to determine whether acceleration sensor values with respect to a bottom, a front, and a side directions of a vehicle are greater than an average acceleration sensor values occurred during past time, the average acceleration sensor values corresponding to each of the acceleration sensor values of the bottom, front, and side directions of the vehicle, and
in response to detection of the impact event being greater than all of the average acceleration sensor values in the bottom, front, and side directions of the vehicle,
the processing unit configured to receive the impact sound and to determine whether the impact sound corresponds to a sound which having been previously set and stored, and
in response determination both of acceleration sensor values with respect to the bottom, front, and side directions of the vehicle are greater than the average acceleration sensor values occurred during the past time and the impact sound corresponds to the sound which having been previously set and stored,
the processing unit further configured to selectively store images through the camera unit in a storage unit, the images corresponding to the selectively stored images captured for a previous specific time and a subsequent specific time based on a point of time at which the impact event is generated.

2. The image storage device of claim 1 , further comprising:
verification sounds being stored in the storage unit, the verification sounds comprising a valid sound and invalid sound.

3. The image storage device of claim 2 , wherein the valid sound is a sound related to a vehicle accident, and the invalid sound is a sound not related to a vehicle accident.

4. The image storage device of claim 2 , wherein the processing unit is configured not to store an image captured through the camera unit in the storage unit in response to detection of the impact event through the acceleration sensor unit and in response to determination of the impact sound received through the microphone unit corresponding to the invalid sound.

5. The image storage device of claim 1 , wherein the processing unit determines whether the impact event is a valid impact event based on a difference between a current acceleration sensor value and an average value of previous acceleration sensor values being greater than a predetermined detection setting value.

6. An image storage method based on input sound and acceleration signals, comprising:
comparing differences between current acceleration sensor values received with respect to a bottom, a front, and a side directions respectively of a vehicle, being greater than an average acceleration sensor values occurred during past time the average acceleration sensor values corresponding to each of the acceleration sensor values of the bottom, front, and side directions of the vehicle;
in response to detection of an impact event being greater than all of the average acceleration sensor values in the bottom, front, and side directions of the vehicle,
receiving impact sound and determining whether the impact sound correspond to a sound which having been previously set and stored; and
in response determination both of acceleration sensor values with respect to the bottom, front, and side directions of the vehicle are greater than the average acceleration sensor values occurred during the past time and the impact sound corresponds to the sound which having been previously set and stored,
selectively storing images, the selectively stored images being captured at a previous specific time and a subsequent specific time based on a point of time at which the impact event is generated.

7. The image storage method of claim 6 , further comprising:
detecting a background noise from an input sound;
extracting a feature point from the input sound excepting the background noise detected; and
classifying the input sound from which the feature point has been extracted into a valid sound and an invalid sound and storing the valid sound and the invalid sound as the verification sounds.

8. The image storage method of claim 7 , wherein the detecting a background noise from the impact sound further comprises:
calculating log energy of the input sound having a frame unit of a length of 30 ms and the impact sound, and detecting the background noise.

9. The image storage method of claim 7 , wherein in the extracting a feature point from the impact sound, a Mel frequency cepstrum coefficient is used to extract the feature points of the input sound and the impact sound excepting the background noise.

10. The image storage method of claim 7 , wherein the-classifying the input sound from which the feature point has been extracted into the valid sound and the invalid sound is performed using a Gaussian mixture model.

11. The image storage method of claim 6 , wherein comparing the impact sound according to the impact event with previously set and stored verification sounds further comprises:
detecting a background noise from the impact sound;
extracting a feature point from the impact sound excepting the background noise;
comparing the feature point of the impact sound extracted in the feature point extraction step with the verification sound; and
determining whether the impact sound is a valid sound or an invalid sound of the verification sound.

12. The image storage method of claim 11 , wherein the comparing the feature point of the impact sound extracted in the feature point extraction step with the verification sound further comprises:
calculating a maximum likelihood value of the impact sound from which the feature point has been extracted, and comparing the maximum likelihood value with the verification sound.

13. The image storage method of claim 11 , wherein the detecting a background noise from the impact sound comprises:
calculating log energy of the input sound having a frame unit of a length of 30 ms and the impact sound, and detecting the background noise.

14. The image storage method of claim 11 , wherein the extracting a feature point from the impact sound excepting the background noise comprises:
a Mel frequency cepstrum coefficient is used to extract the feature points of the input sound and the impact sound other than the background noise.

15. A non-transitory recording medium comprising a computer executable program, which when executed by a processor, instructs the processor to perform the following steps:
comparing differences between current acceleration sensor values received with respect to a bottom, a front, and a side directions respectively of a vehicle, being greater than an average acceleration sensor values occurred during past time the average acceleration sensor values corresponding to each of the acceleration sensor values of the bottom, front, and side directions of the vehicle;
in response to detection of an impact event being greater than all of the average acceleration sensor values in the bottom, front, and side directions of the vehicle,
receiving impact sound and determining whether the impact sound correspond to a sound which having been previously set and stored; and
in response determination both of ac

Assignees

• Ct Integrated Smart Sensors Found

Inventors

• Yang Jin Young
• Park Ji Hoon
• Lim Jin Yeon
• Kyung Chong Min

Classifications

• B60R25/302 Primary

  using recording means, e.g. black box · CPC title

• B60R25/305

  using a camera · CPC title

• B60R21/0136

  responsive to actual contact with an obstacle {, e.g. to vehicle deformation, bumper displacement or bumper velocity relative to the vehicle (monitoring crash strength in combination with passenger seat detection B60R21/01558)} · CPC title

• B60R25/32

  of vehicle dynamic parameters, e.g. speed or acceleration · CPC title

• G07C5/0866 Primary

  the electronic data carrier being a digital video recorder in combination with video camera · CPC title