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

Training method of classifier, image detection method and respective system

A training method and strong classifier technology, applied in the field of image processing, can solve the problems of high classifier training cost, high classifier image detection error rate, and low classification training accuracy.

Active Publication Date: 2016-03-16
ZHANGJIAGANG KANGDE XIN OPTRONICS MATERIAL
View PDF3 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0014] The invention provides a classifier training method, image detection method and respective systems, which are used to solve the problem of high training cost of classifiers in the prior art, low classification training accuracy, and the use of classifier image detection in the prior art. high error rate

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
  • Training method of classifier, image detection method and respective system
  • Training method of classifier, image detection method and respective system
  • Training method of classifier, image detection method and respective system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0045] Such as figure 1 As shown, the present invention provides a training method for a strong classifier. The training method is mainly performed by a training system. The training system is installed in computer equipment. The computer device can also be used for bad block detection based on 3D left and right views. The training samples received by the training system are pre-selected according to the needs of image detection. Each sample contains at least one type of eigenvalues. The feature value types include but are not limited to: difference feature type, SAD feature type, maximum difference feature type, and the like.

[0046] The purpose of the training method is to train a strong classifier with the smallest error probability for a single classification. In order to effectively classify the difference estimation blocks in the image during image detection.

[0047] The training method includes steps S11, S12 and S13. Wherein, the strong classifier is composed ...

Embodiment 2

[0085] In order to train classifiers that can be used for image detection, the present invention also provides a training method for cascade classifiers. Such as Figure 5 shown. The cascaded classifier is composed of several stages of strong classifiers as described in any one of the first embodiment in series.

[0086] Before training the above-mentioned cascade classifier, the training system can use various feature extraction methods to extract the feature values ​​of the sample images (referred to as samples), and after each sample is formed into a sample space, the training system will use the preset samples Each sample in the space is filtered one by one through each order of strong classification.

[0087] Specifically, the training system trains the weak classifiers in the current strong classifier according to the samples received by the current strong classifier, and classifies the weak classifiers in the current strong classifier, The samples in the category wit...

Embodiment 3

[0092] The invention also provides an image detection method, which is used in a detection system of 3D left and right views. Here, the detection system is used to determine whether the corresponding image region is valid by detecting the feature values ​​estimated by its own depth estimation module during the 3D image conversion process. The detection system is preset with cascaded strong classifiers trained through the aforementioned training steps. Wherein, at least two strong classifiers in the cascaded classifier classify the received difference estimation blocks into different classes according to their biased classes, such as Image 6 shown. The image detection method is specifically as follows:

[0093] In step S21, the detection system acquires feature values ​​corresponding to a plurality of difference estimation blocks. Wherein, the feature value includes but not limited to: Disparity feature (Disparity), SAD feature, maximum difference difference feature, and th...

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 invention provides a training method of classifier, an image detection method and a respective system. The training method is used for training cascading strong classifiers. All strong classifiers can be trained in the following steps of (1) initializing sample weight of all samples according to received numbers of the to-be-trained samples; (2) inputting obtained characteristic values and weight of the samples to a weak classifier for classification trainings so as to minimize the error rate in the weak classifier; (3) based on proportion of bias quantity, updating weight of all samples of the next stage weak classifier according to training results of the current weak classifier; (4) repeating steps (2) and (3) until the last stage of weak classifier is trained; and removing samples in the minimal error category classified by the current stage weak classifier, and inputting other parts into the next stage of weak classifier until the last stage of weak classifier is trained. According to the invention, the trained cascading strong classifiers are used for classifying obtained difference evaluation blocks. Problems of low classification accuracy rate and high training cost are solved.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a classifier training method, an image detection method and respective systems. Background technique [0002] Depth estimation problem is the core problem to achieve 3D automatic conversion. 3D automatic conversion refers to converting the traditional 3D left-right image format into a 2D+Z (depth) format that can be used for multi-angle 3D image generation. [0003] Generally speaking, 3D automatic conversion includes a depth estimation module and a depth enhancement module. Among them, the depth estimation module generates a coarse difference map (low resolution) based on the input left and right maps. The coarse difference map (low resolution) output by it is in units of blocks, and each block includes N*N pixels. [0004] The 3D depth enhancement module completes operations such as classification, filtering, and interpolation based on the rough difference map and possible ot...

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): G06K9/62
CPCG06T7/0002G06F18/2113G06F18/24
Inventor 于炀
Owner ZHANGJIAGANG KANGDE XIN OPTRONICS MATERIAL
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