Adaptive Clustering Segmentation Algorithm for Depth Image
An adaptive clustering and depth image technology, applied in image analysis, image data processing, computing and other directions, can solve the problems of target segmentation errors, poor applicability, and difficulty in algorithm convergence iterations, achieving fewer iterations and faster convergence. , with real-time effect
Active Publication Date: 2018-02-23
JINLING INST OF TECH
View PDF4 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The first method is given manually based on experience, and the threshold is set by estimating the distance between the foreground target and the acquisition device, which has poor applicability and may easily cause target segmentation errors; the second method often specifies the number of categories in advance, and then uses algorithms such as K-means Clustering can obtain information such as the category center, but it is easy to make the algorithm difficult to converge or the number of iterations is too large, or even the classification error
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0022] Such as figure 1 As shown, the collected depth image is filtered and preprocessed to extract the histogram parameters of the depth image; the number of categories is determined according to the histogram data and the initial class center point is selected, and the clustering algorithm based on K-means is used for clustering, and the iteration is completed After that, information such as the category center and range can be obtained; according to the distance characteristics of the target, the segmentation threshold can be obtained.
[0023] The present invention adopts following technical scheme:
[0024] An adaptive clustering and segmentation algorithm for depth images, the steps are as follows:
[0025] Step 1: Establish a depth image sample library;
[0026] Using Microsoft's somatosensory peripheral 3D camera Kinect to collect depth images, the scene is not limited, the foreground target is mainly the operator or the operator's face, hand, arm, torso;
[0027] A...
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
Login to View More Abstract
The invention provides an adaptive clustering and segmentation algorithm of a depth image, and relates to the technical field of digital image processing. The method includes: obtaining a depth image, performing preprocessing such as filtering on the collected image samples, extracting a histogram of the depth image, clustering the histogram data by using an improved adaptive K-means algorithm, and obtaining a category center and a category label, etc. information, as a threshold for segmenting depth images. The invention can adaptively determine the number of objects in the depth image, enhance the convergence of the K-means algorithm, and reduce the number of iterations of the K-means algorithm, thereby quickly and effectively segmenting the depth image.
Description
technical field [0001] The invention relates to a depth image self-adaptive clustering and segmentation algorithm based on K-means (K-means), and belongs to the technical field of digital image processing. Background technique [0002] A depth image refers to an image with depth information. The pixel value in the image reflects the distance of the target depth of field, and reflects the real-world distance of the pixel from the acquisition device. [0003] Because it contains the depth information of the target, the depth image makes up for the shortcomings of traditional image acquisition equipment that can only acquire two-dimensional images, and meets the requirements of machine vision for three-dimensional object recognition. Since the depth image has the advantages of not being affected by the direction of the light source and the surface characteristics of the object, there is no shadow, and it is not affected by signals with similar colors, scholars at home and abroa...
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
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/62G06T7/136G06T7/194
CPCG06F18/23213
Inventor 胡勇
Owner JINLING INST OF TECH