Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Image classification and processing method based on different illuminances

A grayscale image and image technology, applied in the field of image processing, can solve problems such as large error in results, dark picture quality, and decreased contrast of surveillance video

Active Publication Date: 2016-11-30
BEIJING UNIV OF TECH
View PDF9 Cites 49 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to propose a classification and processing method based on images with different illuminances, aiming at problems such as large result errors existing in the existing image classification methods and problems such as decreased contrast and darker picture quality in surveillance videos.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Image classification and processing method based on different illuminances
  • Image classification and processing method based on different illuminances
  • Image classification and processing method based on different illuminances

Examples

Experimental program
Comparison scheme
Effect test

Embodiment approach 1

[0166] Embodiment 1, using the SVM algorithm to construct a classifier. Introduce the radial basis (RBF) kernel function of SVM classifier principle, construct the SVM learning training based on RBF kernel function, then the concrete realization of step S32 is as follows:

[0167] Step S321, set the feature set label based on the image feature set obtained in step S31, and set the feature set label of one type of image to 1 as the first image type of classification; the feature set label of the remaining image types is -1 .

[0168] Step S322, introducing the Radial Basis (RBF) kernel function of the principle of the SVM classifier, collecting and training the feature set based on the SVM learning and training of the RBF kernel function, and initially forming a training model.

[0169] Step S323, optimize the parameters in the above training model by means of cross-validation, select the optimal value, and obtain the first SVM classification prediction model.

[0170] Step S...

Embodiment approach 2

[0172] Embodiment 2, using the k-means algorithm to construct an image classifier. Then the concrete realization of step S32 is as follows:

[0173] In step S321, four observation points are randomly selected from the feature sets of the four types of illumination images obtained in step S31 as the data centers of the four types of clusters.

[0174] Step S322, respectively calculate the Euclidean distances from the remaining characteristic data to the four data centers, and assign these characteristic data to the clusters with the closest Euclidean distances to the data centers.

[0175] Step S323, according to the clustering result, calculate the arithmetic mean of all characteristic data in the 4 clusters, and use it as the new data center of each cluster.

[0176] Step S324, re-clustering all feature data according to the new data center.

[0177] Step S325, repeating step S324 until the clustering result no longer changes, forming a classification prediction classifier....

Embodiment approach 3

[0191] Embodiment 3, using a neural network algorithm to construct an image classifier. Then the concrete realization of step S32 is as follows:

[0192] Step S321, initialize the network model and network parameters.

[0193] Step S322, input training samples according to the image feature data obtained in step S31.

[0194] Step S323, calculating the input value and output value of each layer during the forward propagation of the BP neural network.

[0195] Step S324, calculating the output error of each layer of neurons according to the result of step S323.

[0196] Step S325, error backpropagation, using the gradient descent method to adjust the weights and thresholds of each layer.

[0197] Step S326, calculating the global error.

[0198] Step S327, judging whether the training termination condition is satisfied. When the global error reaches the preset accuracy or the number of learning times is greater than the set maximum number of times or all samples are traine...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The present invention discloses an image classification and processing method based on different illuminances. The realization method mainly comprises the steps: collecting video images with different illuminances as training sample images; extracting the features of the sample images as experiment data; performing learning training of the feature data, and forming a classifier; realizing the classification of unknown images with different illuminances; outputting unknown image classification results; automatically determining whether the illuminance processing needs to be performed or not according to the unknown image classification results; performing processing of the unknown images requiring illuminance processing; and outputting the final result images. The classification of images with different illuminances is realized, the problem is solved that the threshold determination error is big, the classification precision is greatly improved, and the work efficiency of the monitoring video image processing technology is improved.

Description

technical field [0001] The invention provides an image classification and processing method based on different illuminance, which belongs to the field of image processing. This method extracts image features by analyzing the characteristics of different illumination images, constructs an image classifier according to the features, realizes automatic classification and judgment of different illumination images, and realizes the image quality problems such as low contrast and dark image quality according to the automatic judgment results Perform illumination processing, and finally output a clear image. Background technique [0002] In recent years, digital surveillance systems have been widely used in public places and become an indispensable part of public safety. However, in the face of different weather and light changes during the day, the contrast and image quality of the video acquired by the digital surveillance system will be affected. For example, the surveillance v...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/46G06K9/62
CPCG06V10/443G06F18/23213G06F18/2411
Inventor 张文利李红璐李会宾张露
Owner BEIJING UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com