Image Approximation Method Based on Extended Smoothlets Transformation

Abstract

The invention discloses an image approximation based on Extended Smoothlets (ExSmoothlets) transformation. The Smoothlets transformation is to realize the description of the transition zone along the horizontal or vertical direction by using translational edge curves. However, in practical applications, many edges cannot be obtained by horizontal or vertical translation. At this time, it is difficult for Smoothlets to accurately express the image. The invention realizes the ExSmoothlets transformation by establishing a more generalized curve model and defining a transition zone direction function. ExSmoothlets translates the curve according to the direction function, and the parameters of the transition zone are determined by the distance from the point inside the transition zone to the edge curve. Compared with the Smoothlets transformation, the invention can better express the image and improve the image approximation accuracy.

Description

Technical field:

[0001] The invention is used in the technical field of image processing, and specifically relates to the field of image approximation. Background technique:

[0002] Since the geometric features of an image are the most important information from the perspective of the human visual system, the study of geometric multi-scale methods for image representation is becoming more and more common, and these multi-scale methods are roughly divided into two categories. One is frame-based non-adaptive computing methods, such as: ridgelets, curvelets, contourlets, shearlets, directionlets. The other is adaptive approximation methods, such as wedgelets, beamlets, second order wedgelets, platelets, surflets, smoothlets. In recent years, basis-based adaptive algorithms have also emerged, such as brushlets, bandelets, grouplets, and tetralets. These "X-lets" are called ultrawavelets, and more and more "X-lets" are defined.

[0003] Smoothlets is an ultrawavelet transform...

Examples

Embodiment Construction

[0052] All features disclosed in this specification, or steps in all methods or processes disclosed, may be combined in any manner, except for mutually exclusive features and / or steps.

[0053] The present invention mainly adopts the method of computer simulation for verification, and all steps and conclusions are verified correctly on MATLAB-R2015a. The specific implementation steps are as follows:

[0054] Step 1. Block the image

[0055] Input a grayscale image, the image size is 256×256. The image is layered, the maximum layer J=8, because the size of the block image is 1×1 and 2×2 contains too few pixel values, it is meaningless to approximate them, so the size of the smallest image block is set to 4 . For layer j, the tiled image size is 2 j ×2 j (j=2, 3, ..., 8), and divide 256×256 into 7 layers, the size of the first layer image block is 256×256, and record the coordinates of the upper left corner of the image, namely (1, 1); The size of the second-layer image bl...

Technical field:

[0001] The invention is used in the technical field of image processing, and specifically relates to the field of image approximation. Background technique:

[0002] Since the geometric features of an image are the most important information from the perspective of the human visual system, the study of geometric multi-scale methods for image representation is becoming more and more common, and these multi-scale methods are roughly divided into two categories. One is frame-based non-adaptive computing methods, such as: ridgelets, curvelets, contourlets, shearlets, directionlets. The other is adaptive approximation methods, such as wedgelets, beamlets, second order wedgelets, platelets, surflets, smoothlets. In recent years, basis-based adaptive algorithms have also emerged, such as brushlets, bandelets, grouplets, and tetralets. These "X-lets" are called ultrawavelets, and more and more "X-lets" are defined.

[0003] Smoothlets is an ultrawavelet transform...

Examples

Embodiment Construction

[0052] All features disclosed in this specification, or steps in all methods or processes disclosed, may be combined in any manner, except for mutually exclusive features and / or steps.

[0053] The present invention mainly adopts the method of computer simulation for verification, and all steps and conclusions are verified correctly on MATLAB-R2015a. The specific implementation steps are as follows:

[0054] Step 1. Block the image

[0055] Input a grayscale image, the image size is 256×256. The image is layered, the maximum layer J=8, because the size of the block image is 1×1 and 2×2 contains too few pixel values, it is meaningless to approximate them, so the size of the smallest image block is set to 4 . For layer j, the tiled image size is 2 j ×2 j (j=2, 3, ..., 8), and divide 256×256 into 7 layers, the size of the first layer image block is 256×256, and record the coordinates of the upper left corner of the image, namely (1, 1); The size of the second-layer image bl...

Images