CN103679708A - Annular LBP texture generating method - Google Patents

Abstract

The invention relates to an annular LBP texture generating method. The method includes: extending into a 2*2 four-point mode by using one specific pixel point as the left upper corner, and performing weighted coding according to the clockwise side of the four pixel points to serve as the feature value of the specific pixel point. The annular LBP texture generating method is good in extracting effect, fast, short in operation time, and wide in application range, and no information redundancy is generated.

Description

A kind of annular LBP Texture Generating Approach

Technical field

The present invention relates to image texture method, be specifically related to a kind of annular LBP Texture Generating Approach.

Background technology

Development along with remote sensing images analysis, recognition of face and the identification of target following isotype and machine vision, texture analysis method is being brought into play increasing effect, yet due to reasons such as illumination reason, background complexity, target occlusions, make the pattern-recognition in above-mentioned field become difficult.The a large amount of texture analysis methods that occur in recent years, real-time and accuracy all depend on related image.

Local binary pattern (LBP) is to identify a kind of more successful texture characteristic extracting method in isotype identification field remote sensing image processing, the mankind, LBP obtains by the poor of the pixel value of computing center's pixel and neighborhood pixel, there is following problem: one, the method has certain information redundancy, the pixel of part has participated in double counting, therefore there is higher computation complexity, be unfavorable for processing in real time; Its two, the method fails to consider the relation between neighborhood pixel, and magnitude relationship between neighborhood pixel has also comprised a large amount of information, so classic method has reduced the separating capacity of LBP to the textural characteristics of true picture virtually.

Therefore, the New Ring-like Type LBP Texture Generating Approach that the present invention proposes is very meaningful, also can play positive role to researchs such as similar pattern-recognition, machine vision.

Summary of the invention

Goal of the invention: in order to overcome the deficiencies in the prior art, the invention provides a kind of based on 4 annulars and the annular LBP Texture Generating Approach of exquisite pattern, the method can effectively represent the local grain information of remote sensing image and mankind's image, and computation complexity is low.

Technical scheme: for solving the problems of the technologies described above, a kind of annular LBP Texture Generating Approach of the present invention, comprises the following steps:

(1) set up 4 annular binary patterns: a pixel and this specific pixel are put a bottom-right pixel to choose a pixel that certain specific pixel point in image, this specific pixel point the right are adjacent, this specific pixel point below, these four pixels form 2 * 2 block of pixels, give this four corresponding weights of pixel, from specific pixel point, start by arranged clockwise, the weight size of each pixel is s=2 ^2i-1, i=1,2,3,4;

Wherein, this weight can embody the difference of textural characteristics value under various arrangement modes as far as possible, and can be distributed between 0 to 255 in mode preferably, is so relatively conducive to storage and the demonstration of the 8bit image of image;

(2), from specific pixel point, the difference between the clockwise next pixel of this pixel and this pixel relatively successively, if difference is more than or equal to 0, the LBP that this pixel is corresponding is encoded to 1, if difference is less than 0, the LBP that this pixel is corresponding is encoded to 0

$s({\mathrm{im}}_i,{\mathrm{im}}_{i+1})=\left\{\begin{array}{cc}1& {\mathrm{im}}_i\&GreaterEqual;{\mathrm{im}}_{i+1}\\ 0& {\mathrm{im}}_i<{\mathrm{im}}_{i+1}\end{array}\right.,$

Wherein, im _irefer to i pixel, im _i+1refer to the clockwise next pixel of i pixel, s (im _i, im _i+1) refer to that the LBP of i pixel encodes;

(3) the annular LBP coding of four pixels of gained in step (2) is multiplied by after corresponding weight, by four value summations of gained, can obtains the annular textural characteristics value RBP of specific pixel point,

$\mathrm{RBP}=\underset{i=1}{\overset{4}{\mathrm{\&Sigma;}}}{2}^{2i-1}s({\mathrm{im}}_i,{\mathrm{im}}_{i+1}),$

Wherein, 2 ^2i-1be the weight of i pixel, s (im _i, im _i+1) refer to the coding of i pixel.

Further, travel through whole image, each pixel in image is carried out to the operation of step (1), (2) and (3), can obtain the whole annular LBP characteristic image of image.

Beneficial effect: a kind of annular LBP Texture Generating Approach of the present invention proposes based on gray level image, extraction effect is good, speed is fast, the arithmetic speed time is short, can not produce information redundancy, and the present invention has taken into full account the relation between neighbor pixel, can not repeat to extract, hormone complexity of the present invention is low simultaneously, is conducive to process in real time.Compare with traditional LBP Texture Generating Approach, the invention enables between different classes of ground object difference larger, and the enhancing of identical category interior of articles homogeney, the present invention can be applied to a plurality of fields such as remote sensing image processing, recognition of face, target following in addition, and applicability is wide.

Accompanying drawing explanation

Fig. 1 utilizes the present invention to carry out the schematic diagram that annular LBP textural characteristics generates;

Fig. 2 utilizes the present invention to carry out the computation process schematic diagram that annular LBP textural characteristics calculates.

Embodiment

Below in conjunction with accompanying drawing, the present invention is further described.

As depicted in figs. 1 and 2, a kind of annular LBP Texture Generating Approach of the present invention, comprises the following steps:

(1) set up 4 annular binary patterns: a pixel and this specific pixel are put a bottom-right pixel to choose a pixel that certain specific pixel point in image, this specific pixel point the right are adjacent, this specific pixel point below, these four pixels form 2 * 2 block of pixels, give this four corresponding weights of pixel, from specific pixel point, start by arranged clockwise, the weight size of each pixel is s=2 ^2i-1, i=1,2,3,4;

(2), from specific pixel point, the difference between the clockwise next pixel of this pixel and this pixel relatively successively, if difference is more than or equal to 0, the LBP that this pixel is corresponding is encoded to 1, if difference is less than 0, the LBP that this pixel is corresponding is encoded to 0

$s({\mathrm{im}}_i,{\mathrm{im}}_{i+1})=\left\{\begin{array}{cc}1& {\mathrm{im}}_i\&GreaterEqual;{\mathrm{im}}_{i+1}\\ 0& {\mathrm{im}}_i<{\mathrm{im}}_{i+1}\end{array}\right.,$

Wherein, im _irefer to i pixel, im _i+1refer to the clockwise next pixel of i pixel, s (im _i, im _i+1) refer to that the LBP of i pixel encodes;

(3) the annular LBP coding of four pixels of gained in step (2) is multiplied by after corresponding weight, by four value summations of gained, can obtains the annular textural characteristics value RBP of specific pixel point,

$\mathrm{RBP}=\underset{i=1}{\overset{4}{\mathrm{\&Sigma;}}}{2}^{2i-1}s({\mathrm{im}}_i,{\mathrm{im}}_{i+1}),$

Wherein, 2 ^2i-1be the weight of i pixel, s (im _i, im _i+1) refer to the coding of i pixel.

Finally each pixel in image is carried out the operation of step (1), (2) and (3), obtain the whole annular LBP characteristic image of image.

The above is only the preferred embodiment of the present invention; be noted that for those skilled in the art; under the premise without departing from the principles of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (2)

1. an annular LBP Texture Generating Approach, is characterized in that comprising the following steps:

(1) set up 4 annular binary patterns: a pixel and this specific pixel are put a bottom-right pixel to choose a pixel that certain specific pixel point in image, this specific pixel point the right are adjacent, this specific pixel point below, these four pixels form 2 * 2 block of pixels, give this four corresponding weights of pixel, from specific pixel point, start by arranged clockwise, the weight size of each pixel is s=2 ^2i-1, i=1,2,3,4;

(2), from specific pixel point, the difference between the clockwise next pixel of this pixel and this pixel relatively successively, if difference is more than or equal to 0, the LBP that this pixel is corresponding is encoded to 1, if difference is less than 0, the LBP that this pixel is corresponding is encoded to 0

$s({\mathrm{im}}_i,{\mathrm{im}}_{i+1})=\left\{\begin{array}{cc}1& {\mathrm{im}}_i\&GreaterEqual;{\mathrm{im}}_{i+1}\\ 0& {\mathrm{im}}_i<{\mathrm{im}}_{i+1}\end{array}\right.,$

Wherein, im _irefer to i pixel, im _i+1refer to the clockwise next pixel of i pixel, s (im _i, im _i+1) refer to that the LBP of i pixel encodes;

(3) the annular LBP coding of four pixels of gained in step (2) is multiplied by after corresponding weight, by four value summations of gained, can obtains the annular textural characteristics value RBP of specific pixel point,

$\mathrm{RBP}=\underset{i=1}{\overset{4}{\mathrm{\&Sigma;}}}{2}^{2i-1}s({\mathrm{im}}_i,{\mathrm{im}}_{i+1}),$

Wherein, 2 ^2i-1be the weight of i pixel, s (im _i, im _i+1) refer to the coding of i pixel.

2. annular LBP Texture Generating Approach according to claim 1, is characterized in that: each pixel in image is carried out to the operation of step (1), (2) and (3), obtain the whole annular LBP characteristic image of image.