Device for processing image and operating method thereof

Patent metadata

Field	Value
Publication number	US-12567166-B2
Application number	US-202218081195-A
Country	US
Kind code	B2
Filing date	Dec 14, 2022
Priority date	Jun 19, 2020
Publication date	Mar 3, 2026
Grant date	Mar 3, 2026

Abstract

A processing device and an operating method of the processing device are provided. An embodiment of the present disclosure provides the processing device that obtains a reconstructed image of the raw image by compressing a raw image at a preset compression ratio, encoding the compressed raw image, and decoding the encoded image, recognizes, in the raw image, a target object and a first region of interest (ROI) including the target object and extracts a first feature vector from the target object recognized in the raw image, recognizes, in the reconstructed image, the target object and a second ROI including the target object and extracts a second feature vector from the target object recognized in the reconstructed image, determines whether the target object is recognized in the reconstructed image, based on a degree of overlap between the first and second ROIs and a degree of similarity between the first and second feature vectors, and adjusts the preset compression ratio according to whether the target object is recognized.

First claim

What is claimed is:

1 . A method performed by a processing device, the method comprising:
obtaining a raw image from at least one imaging sensor; obtaining a reconstructed image from the raw image by operations of compressing the raw image at a preset compression ratio, encoding the compressed raw image, and decoding the encoded compressed raw image; recognizing, in the raw image, a target object and a first region of interest (ROI) including the target object, and extracting a first feature vector from the target object recognized in the raw image; recognizing, in the reconstructed image, the target object and a second ROI including the target object, and extracting a second feature vector from the target object recognized in the reconstructed image; calculating a degree of overlap between the first ROI and the second ROI based on coordinate value information of each of a first bounding box of the first ROI and a second bounding box of the second ROI; determining, based on the calculated degree of overlap between the first ROI and the second ROI and a degree of similarity between the first feature vector and the second feature vector, whether the target object is recognized in the reconstructed image; and adjusting the preset compression ratio according to whether the target object is recognized.

2 . The method of claim 1 , wherein the determining of whether the target object is recognized comprises:
calculating the degree of similarity between the first feature vector and the second feature vector; comparing the calculated degree of overlap with a preset first threshold and comparing the calculated degree of similarity with a preset second threshold; and determining, based on results of the comparisons, whether the target object is recognized in the reconstructed image.

3 . The method of claim 2 , wherein the adjusting of the preset compression ratio based on the comparing between the calculated degree of overlap with the first threshold or the comparing between the calculated degree of similarity with the second threshold threshold, comprises lowering a value of the preset compression ratio based on (i) the degree of overlap that is less than or equal to the first threshold or (ii) the degree of similarity that is less than or equal to the second threshold.

4 . The method of claim 3 , further comprising obtaining another reconstructed image of the raw image by compressing the raw image at an adjusted preset compression ratio, encoding the compressed raw image, and decoding the encoded image, wherein the adjusted preset compression ratio has the value that is lowered based on (i) the degree of overlap that is less than or equal to the first threshold or (ii) the degree of similarity that is less than or equal to the second threshold.

5 . The method of claim 2 , wherein the adjusting of the preset compression ratio comprises determining the preset compression ratio to obtain the reconstructed image as a final compression ratio based on the calculated degree of overlap exceeds the first threshold or the calculated degree of similarity exceeds the second threshold.

6 . The method of claim 2 , wherein the calculating of the degree of overlap comprises:
obtaining first position coordinate values of the first ROI that is a region where the target object is recognized from the raw image; obtaining second position coordinate values of the second ROI that is a region where the target object is recognized from the reconstructed image; and calculating the degree of overlap, which indicates a degree to which the first ROI and the second ROI overlap, by using the first position coordinate values and the second position coordinate values.

7 . The method of claim 1 , wherein the reconstructed image is obtained by compressing, encoding, and decoding the raw image by using at least one image compression standard, and
wherein the preset compression ratio is determined by a quantization parameter (QP) in the at least one image compression standard.

8 . A processing device comprising:
an imaging sensor configured to obtain a raw image by capturing an image of an object; an encoder comprising:
an encoding module configured to generate encoded image data by compressing the raw image at a preset compression ratio and encoding the compressed raw image into image data, and
a decoding module configured to generate a reconstructed image from the raw image by decoding the encoded compressed raw image data;
a network interface configured to receive the raw image and the reconstructed image; a memory storing a program including one or more instructions; and at least one processor configured to execute the one or more instructions of the program stored in the memory to:
recognize, in the raw image, a target object and a first region of interest (ROI) including the target object, and extract a first feature vector from the target object recognized in the raw image,
recognize, in the reconstructed image, the target object and a second ROI including the target object, and extract a second feature vector from the target object recognized in the reconstructed image,
calculate a degree of overlap between the first ROI and the second ROI based on coordinate value information of each of a first bounding box of the first ROI and a second bounding box of the second ROI,
determine, based on the calculated degree of overlap between the first ROI and the second ROI and a degree of similarity between the first feature vector and the second feature vector, whether the target object is recognized in the reconstructed image, and
adjust the preset compression ratio according to whether the target object is recognized.

9 . The processing device of claim 8 , wherein the at least one processor is further configured to:
calculate the degree of similarity between the first feature vector and the second feature vector, compare the calculated degree of overlap with a preset first threshold and compare the calculated degree of similarity with a preset second threshold, and determine, based on results of the comparisons, whether the target object is recognized in the reconstructed image.

10 . The processing device of claim 9 , wherein the at least one processor is further configured to lower a value of the preset compression ratio based on (i) the degree of overlap that is less than or equal to the first threshold or (ii) the degree of similarity that is less than or equal to the second threshold.

11 . The processing device of claim 10 , wherein the at least one processor is further configured to obtain another reconstructed image from the raw image by compressing the raw image at an adjusted preset compression ratio, encoding the compressed raw image, and decoding the encoded compressed raw image, and
wherein the adjusted preset compression ratio has the value that is lowered based on (i) the degree of overlap that is less than or equal to the first threshold or (ii) the degree of similarity that is less than or equal to the second threshold.

12 . The processing device of claim 9 , wherein the at least one processor is further configured to determine to obtain the reconstructed image as a final compression ratio based on the calculated degree of overlap exceeds the first threshold or the calculated degree of similarity exceeds the second threshold.

13 . The processing device of claim 8 , wherein the at least one processor is further configured to:
obtain first position coordinate values of the first ROI that is a region where the target object is recognized from the raw image, obtain second position coordinate values of the second ROI that is a r

Assignees

• Samsung Electronics Co Ltd

Inventors

• Kim Dongchan
• Hwang Jinyoung
• Piao Yinji

Classifications

• G06T9/00

  Image coding (bandwidth or redundancy reduction for static pictures H04N1/41; coding or decoding of static colour picture signals H04N1/64; methods or arrangements for coding, decoding, compressing or decompressing digital video signals H04N19/00) · CPC title

• G06V2201/07

  Target detection · CPC title

• G06V10/44

  Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components · CPC title

• G06V10/761

  Proximity, similarity or dissimilarity measures · CPC title

• G06V10/25

  Determination of region of interest [ROI] or a volume of interest [VOI] · CPC title