A computer digital image fast processing system

Through the Dssd_Inception_V4_Objects365 model and Hadoop distributed processing technology, the problems of slow speed and poor quality of traditional image processing systems are solved, and the fast, accurate processing and efficient analysis of images are achieved, improving image quality and user experience.

CN112396552BActive Publication Date: 2024-11-22JILIN UNIVERSITY
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202011576454.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2020-12-28
Publication Date
2024-11-22
Estimated Expiration
2040-12-28

AI Technical Summary

Technical Problem

Traditional computer digital image processing systems operate slowly, have large errors in image processing results, and poor quality.

Method used

The Dssd_Inception_V4_Objects365 model is used for image segmentation, matching the processing algorithm groups of different image blocks, and distributed processing is used using Hadoop, combining image preprocessing, fusion modules and hyperlink marking points for image analysis.

Benefits of technology

It realizes fast and accurate image processing, improves image quality and accuracy of analysis results, improves processing efficiency and facilitates users to obtain information.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112396552B_ABST
    Figure CN112396552B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image processing, and in particular to a computer digital image rapid processing system, comprising: an image segmentation module, for realizing the recognition of objects contained in an image based on a Dssd_Inception_V4_Objects365 model, and dividing the image into a plurality of image blocks according to the recognition result; an image processing module, for matching a corresponding image processing algorithm group for each image block according to the recognition result of the Dssd_Inception_V4_Objects365 model, and realizing the processing of the image block by running the image processing algorithm group based on Hadoop; and an image fusion module, for realizing the fusion processing of the image blocks. The present invention can realize the regional targeted processing of image blocks, greatly improve the quality of the obtained image and the accuracy of the obtained image analysis results, and improve the image processing efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of image processing, and in particular to a computer digital image rapid processing system. Background Art

[0002] Digital image processing, also known as computer image processing, is a method and technology for removing noise, enhancing, restoring, segmenting, extracting features, etc. from images through computers, usually achieved through specific image processing algorithms.

[0003] At present, traditional computer data image processing systems generally have problems such as slow operation speed, large errors in image processing results, and poor quality. Summary of the invention

[0004] To solve the above problems, the present invention provides a computer digital image rapid processing system, which can realize rapid and accurate processing of digital images.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A computer digital image rapid processing system, comprising:

[0007] Image segmentation module, used to recognize objects in images based on the Dssd_Inception_V4_Objects365 model, and segment the image into several image blocks according to the recognition results;

[0008] An image processing module is used to match a corresponding image processing algorithm group for each image block according to the recognition result of the Dssd_Inception_V4_Objects365 model, and run the image processing algorithm group based on Hadoop to realize the processing of the image block;

[0009] The image fusion module is used to realize the fusion processing of image blocks.

[0010] Furthermore, the plurality of image blocks are composed of a background block and a plurality of object blocks, and each object block includes only one category of objects.

[0011] Furthermore, the background block and different object blocks correspond to different image processing algorithm groups respectively.

[0012] Furthermore, it also includes:

[0013] The image preprocessing module is used to generate a grayscale image of the received image according to the edge strength of the pixel points of the image, and to perform sharpening processing on the image based on the grayscale image to obtain a processed image.

[0014] Furthermore, the image fusion module realizes the fusion of the image blocks based on the position coordinates of the image blocks in the original image.

[0015] Furthermore, during image analysis, each object block carries a hyperlink mark point corresponding to its analysis result, and the background block carries a hyperlink mark point corresponding to the image (whole image) analysis result.

[0016] Furthermore, during image analysis, the image processing module first runs the image processing algorithm group based on Hadoop to process each image block, and then enters the analysis results of each image block into a preset image processing model to obtain the corresponding image analysis results.

[0017] Furthermore, the image processing algorithm group contains a plurality of image processing algorithms, and each image processing algorithm is provided with an independent number. Hadoop runs the image processing algorithms in sequence according to the number order to realize the processing of the image blocks.

[0018] The present invention has the following beneficial effects:

[0019] 1) The Dssd_Inception_V4_Objects365 model is used to recognize objects in images, and the images are divided into several image blocks according to the recognition results. Different image processing algorithm groups are matched for the background blocks and different object blocks, so that the image blocks can be processed in different regions, which greatly improves the quality of the obtained images and the accuracy of the obtained image analysis results.

[0020] 2) The image processing algorithm group is run based on Hadoop to process each image block, thereby realizing distributed loading and analysis of image blocks, greatly improving image processing efficiency, and thus realizing rapid image processing.

[0021] 3) The image analysis results are fed back by means of hyperlinked marking points, and the analysis results are displayed based on the fused overall image, so that users can obtain the information contained in the current image at a glance, which greatly facilitates the user's subsequent use. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 This is a system block diagram of a computer digital image rapid processing system according to embodiment 1 of the present invention.

[0023] Figure 2 The following is a flowchart of an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0025] like Figure 1 As shown, an embodiment of the present invention provides a computer digital image rapid processing system, comprising:

[0026] An image preprocessing module, used to generate a grayscale image of the received image according to the edge strength of the pixel points of the image, and to perform sharpening processing on the image based on the grayscale image to obtain a processed image;

[0027] The image segmentation module is used to recognize objects in an image based on the Dssd_Inception_V4_Objects365 model, and to segment the image into several image blocks according to the recognition results; the several image blocks are composed of a background block and several object blocks, and each object block contains only one category of objects;

[0028] An image processing module is used to match a corresponding image processing algorithm group for each image block according to the recognition result of the Dssd_Inception_V4 Objects365 model, and run the image processing algorithm group based on Hadoop to realize the processing of the image block;

[0029] The image fusion module is used to realize the fusion processing of image blocks.

[0030] In this embodiment, the Dssd_Inception_V4_ Objects365 model adopts the dssd target detection algorithm, pre-trains the Inception_V4 deep neural network with the Objects365 data set, and then trains the model with the previously prepared data set, fine-tunes the various parameters in the deep neural network, and finally is used to implement a target detection model for detecting different objects (such as people, buildings, parts, etc.) in the image;

[0031] In this embodiment, the background block and different object blocks correspond to different image processing algorithm groups respectively.

[0032] In this embodiment, the image preprocessing module implements image preprocessing through the following steps:

[0033] Performing a dilation operation and / or a Gaussian blur operation on the grayscale image to obtain an intermediate image A;

[0034] Performing an erosion operation on the intermediate image A to obtain an intermediate image A1;

[0035] The image is sharpened based on the intermediate image A1.

[0036] In this embodiment, the image processing algorithm group includes a plurality of image processing algorithms, and each image processing algorithm is independently numbered. Hadoop runs the image processing algorithms in sequence according to the numbering sequence to process the image blocks.

[0037] In this embodiment, the image fusion module realizes the fusion of image blocks based on the position coordinates of the image blocks in the original image.

[0038] It is worth noting that when analyzing an image, each object block carries a hyperlink marker point corresponding to its analysis result, and the background block carries a hyperlink marker point corresponding to the image (whole image) analysis result.

[0039] It is worth noting that during image analysis, the image processing module first runs the image processing algorithm group based on Hadoop to process each image block, and then enters the analysis results of each image block into the preset image processing model to obtain the corresponding image analysis results.

[0040] like Figure 2 As shown, this specific implementation includes the following steps:

[0041] S1. Generate a grayscale image of the received image according to the edge strength of the pixel points of the received image, and perform sharpening processing on the image based on the grayscale image to obtain a processed image;

[0042] S2. Recognize the objects in the image based on the Dssd_Inception_V4_ Objects365 model, and divide the image into several image blocks according to the recognition results; the several image blocks are composed of a background block and several object blocks, and each object block contains only one category of objects;

[0043] S3. According to the recognition result of the Dssd_Inception_V4 Objects365 model, a corresponding image processing algorithm group is matched for each image block, and the image processing algorithm group is run based on Hadoop to process the image block;

[0044] S4. Implement fusion processing of image blocks; specifically, implement fusion of image blocks based on position coordinates of the image blocks in the original image.

[0045] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A computer digital image rapid processing system, characterized in that: include: Image segmentation module, used to recognize objects in images based on the Dssd_Inception_V4_Objects365 model, and segment the image into several image blocks according to the recognition results; An image processing module is used to match a corresponding image processing algorithm group for each image block according to the recognition result of the Dssd_Inception_V4_Objects365 model, and run the image processing algorithm group based on Hadoop to realize the processing of the image block; the Dssd_Inception_V4_Objects365 model adopts the dssd target detection algorithm, pre-trains the Inception_V4 deep neural network with the Objects365 data set, and then trains the model with the previously prepared data set, fine-tunes various parameters in the deep neural network, and finally realizes the target detection model for detecting different objects in the image; The image fusion module is used to realize the fusion processing of image blocks.

2. A computer digital image rapid processing system as claimed in claim 1, characterized in that: The plurality of image blocks are composed of a background block and a plurality of object blocks, and each object block includes only one category of objects.

3. A computer digital image rapid processing system as claimed in claim 2, characterized in that: The background block and different object blocks correspond to different image processing algorithm groups respectively.

4. A computer digital image rapid processing system as claimed in claim 1, characterized in that: Also includes: The image preprocessing module is used to generate a grayscale image of the received image according to the edge strength of the pixel points of the image, and to perform sharpening processing on the image based on the grayscale image to obtain a processed image.

5. A computer digital image rapid processing system as claimed in claim 1, characterized in that: The image fusion module realizes the fusion of image blocks based on the position coordinates of the image blocks in the original image.

6. A computer digital image rapid processing system as claimed in claim 1, characterized in that: During image analysis, each object block carries a hyperlink mark point corresponding to its analysis result, and the background block carries a hyperlink mark point corresponding to the image analysis result.

7. A computer digital image rapid processing system as claimed in claim 6, characterized in that: During image analysis, the image processing module first runs the image processing algorithm group based on Hadoop to process each image block, and then enters the analysis results of each image block into a preset image processing model to obtain the corresponding image analysis results.

8. A computer digital image rapid processing system as claimed in claim 1, characterized in that: The image processing algorithm group contains a plurality of image processing algorithms, and each image processing algorithm is provided with an independent number. Hadoop runs the image processing algorithms in sequence according to the number sequence to realize the processing of the image blocks.

Citation Information

Patent Citations

  • Method and device for image recognition

    CN110458107A

  • Urban landscaping information management system

    CN112070485A

  • Super-saturation current field effect transistor and trans-impedance MOS device

    US20190252382A1