CN106023190B

CN106023190B - A kind of evaluation method of the quasi- burnt degree of auto-focusing image

Info

Publication number: CN106023190B
Application number: CN201610325972.7A
Authority: CN
Inventors: 冯华君; 王烨茹; 徐之海; 李奇; 陈跃庭
Original assignee: Zhejiang University ZJU
Current assignee: Zhejiang University ZJU
Priority date: 2016-05-16
Filing date: 2016-05-16
Publication date: 2019-05-07
Anticipated expiration: 2036-05-16
Also published as: CN106023190A

Abstract

The invention discloses a kind of evaluation methods of the quasi- burnt degree of auto-focusing image, the method that this method uses phase equalization map to combine with gradient map, solve the problems, such as that there are peak values to collapse as focusing evaluation function using phase equalization method, no matter the scene that or brightness relatively low for normal scene or noise changes all has good performance capabilities, practical value is higher, has good unbiasedness, unimodality, sensitivity, stability.

Description

A kind of evaluation method of the quasi- burnt degree of auto-focusing image

Technical field

The present invention relates to a kind of evaluation sides of the quasi- burnt degree of image quality evaluating method more particularly to auto-focusing image Method.

Background technique

With the fast development of digital picture and multimedia technology, various types of optical imaging systems are also more and more, Autofocus Technology plays a crucial role in optical imaging systems.There are mainly two types of Autofocus Technologies, respectively Active focusing and passive type focusing.Actively focusing needs that additional range-measurement system is added in systems, increase system at The complexity of this and system, simultaneously for having glass or having the scene of shelter, ranging will appear problem, therefore, active Focusing technology is at leisure replaced self-action focusing technology.

Passive focusing is to judge defocus degree by the fog-level of image obtained in analysis focus process, to refer to Lead focus process.This Atomatic focusing method is to obtain image definition characteristic value using certain Digital Image Processing algorithm, And motor is controlled according to this characteristic value, to drive eyeglass or imaging sensor to change position, corresponding adjustment is done, until this Until one characteristic value meets a certain condition made an appointment.Therefore, the auto focus evaluation function of superperformance is selected to judge The fog-level of image is the key that problem to guide autofocus system accurately to focus.Due to the weight of focusing evaluation function The evaluation of the property wanted, image definition has become a popular research field.Ideal focusing evaluation function should have unbiased Property, unimodality, sensitivity, stability, in some special cases should also to scene light and shade change and noise have preferably Robustness.Existing auto focus evaluation function is very sensitive to the brightness and contrast of scene, because its principle is to utilize figure The intensity of picture judges defocus degree.And with imaging system in people's lives universal, people like in various environment Lower to use focusing system, the case where focusing system is used in the case where brightness is inconsistent or the lower scene of signal-to-noise ratio, is also more and more. In consideration of it, still thering is the evaluation function of good behaviour performance to become the urgent need of people in such scene.

Summary of the invention

It is an object of the invention to collapse as focusing evaluation function there are peak value for using phase equalization method Problem proposes a kind of evaluation method of quasi- burnt degree of auto-focusing image.

The purpose of the present invention is what is be achieved through the following technical solutions: a kind of evaluation side of the quasi- burnt degree of auto-focusing image Method, method includes the following steps:

(1) image I is obtained；

(2) focusing window region I is chosen from image I_c；

(3) focusing window region I is calculated using log-Gabor small echo_cPhase equalization map PC, each pixel Phase equalization is defined as:

Wherein, m and n respectively indicates the direction scale number of Log-Gabor wavelet filter, A_ji(x, y) and φ_ji(x,y) The amplitude and local phase deviation being illustrated respectively in the logarithmic scale of j-th of i-th of direction.T_iIt is estimating on i-th of direction Count noise, W_i(x, y) is the weighting function on i-th of direction, f_p(u)=(u > 0) u.

Δ φ (x, y)=cos [φ_r(x,y)-φ(x,y)]-|sin[φ_r(x,y)-φ(x,y)]| (2)

φ (x, y) is the local phase weighted mean at the position (x, y), φ_r(x, y) represents local phase at (x, y) R grade Fourier expansion series component at position；

(4) G is composed using the gradient magnitude that gradient calculation method obtains phase equalization map PC:

(5) it sums to all gradient magnitudes in G, as final focusing function evaluation of estimate, gradient magnitude summation is higher, Current image location is closer to focus position.

The beneficial effects of the present invention are: 1. solve scene there are brightness change by phase equalization map PC and make an uproar The problem of in terms of sound, in quasi- burnt position when overcoming phase equalization method as evaluation function further combined with gradient calculation method Setting curve nearby there are problems that collapsing, and finally obtained evaluation function curve has good unbiasedness, unimodality, sensitivity And stability, it is various including the situations such as practical value is higher, can be widely used in including brightness change, and signal-to-noise ratio is low Actual photographed scene.

Detailed description of the invention

Fig. 1 is the flow diagram of inventive method.

Fig. 2 is the schematic diagram of optional 5 width in one group of emulation defocus sequence chart.

Burnt image subject to Fig. 3.

Fig. 4 is the focusing window image chosen.

Fig. 5 is the phase equalization map of focusing window area image.

The gradient magnitude that Fig. 6 is phase equalization map PC composes G.

Fig. 7 is the evaluation function curve obtained using phase equalization method.

Fig. 8 is the evaluation function curve that method obtains in conjunction with gradient using phase equalization.

Fig. 9 is the evaluation function curve obtained using existing method.

Specific embodiment

A kind of evaluation method of the quasi- burnt degree of auto-focusing image of the present invention, this method is using phase equalization map and ladder The method that degree map combines, there are what peak value collapsed to ask when solving using phase equalization method as focusing evaluation function Topic, no matter the scene that or brightness relatively low for normal scene or noise changes all has good performance capabilities, practical valence Value is higher, and application range is wider, has good unbiasedness, unimodality, sensitivity, stability.

It is described in detail with reference to the accompanying drawing with example:

Fig. 1 is the simple process block diagram of the method for the present invention.Below with reference to embodiment, the invention will be further described.

For the present embodiment by taking one group of emulation defocus sequence chart as an example, Fig. 2 is that 5 width figures are arbitrarily chosen in one group of defocus sequence chart As schematic diagram, wherein (a)-(e) is respectively the 1st width, the 9th width, the 17th width, the 25th width, the 33rd width defocus sequence analogous diagram, wherein Burnt figure subject to 17th width.Wherein, it is gradually become smaller from the 1st width~the 17th width defocusing amount, gradually from the 17th width~the 33rd width defocusing amount Become larger.

The present embodiment carries out defocusing amount evaluation to 33 width figures using existing method first, existing since brightness is inconsistent Method can not really reflect the defocusing amount of each figure, as shown in Figure 9.

The present embodiment further uses merely using phase equalization method and method of the invention to above-mentioned 33 width Image carries out evaluating following (by taking the 17th width image as an example):

(1) it is as shown in Figure 3 to obtain image I；

(2) the focusing window region I of 200 × 200 pixel as shown in Figure 4 is chosen from image I_c；

(3) focusing window area image I is calculated using log-Gabor small echo_cPhase equalization map PC, 2D scheme As in, the phase equalization of each pixel is defined as:

Δ φ (x, y)=cos [φ_r(x,y)-φ(x,y)]-|sin[φ_r(x,y)-φ(x,y)]| (2)

φ (x, y) is the local phase weighted mean at the position (x, y), φ_r(x, y) represents local phase at (x, y) R grade Fourier expansion series component at position, phase equalization map PC are as shown in Figure 5；

(4) G is composed using the gradient magnitude that gradient calculation method obtains phase equalization map PC, as shown in fig. 6,

Wherein, G_xWith G_yRespectively indicate the partial derivative of phase equalization map PC in the horizontal and vertical directions；

(5) it sums to all gradient magnitudes in G, as final focusing function evaluation of estimate.

It repeats the above steps and the calculating of clarity evaluation of estimate is carried out to 33 width defocus sequence analogous diagrams of acquisition, and obtain the group The evaluation function curve of defocus sequence focusing window, as shown in Figure 8.It preferably solves as shown in Figure 7 simple using phase Coherence method has that peak value collapses as focusing evaluation function.

The auto focus evaluation function calculation method that the invention proposes a kind of phase equalizations in conjunction with gradient, this method Solve the problems, such as using phase equalization method as focusing evaluation function there are peak values to collapse, no matter for normal scene also It is that the scene that noise is relatively low or brightness changes all has good performance capabilities, practical value is higher, and application range is wider, tool There are good unbiasedness, unimodality, sensitivity, stability.

Claims

1. a kind of evaluation method of the quasi- burnt degree of auto-focusing image, which is characterized in that method includes the following steps:

(1) image I is obtained；

(2) focusing window region I is chosen from image I_c；

(3) focusing window region I is calculated using log-Gabor small echo_cPhase equalization map PC, the phase of each pixel Conformance definition are as follows:

Wherein, m and n respectively indicates the direction scale number of Log-Gabor wavelet filter, A_ji(x, y) and φ_ji(x, y) difference Indicate amplitude and local phase deviation in i-th of j-th of direction logarithmic scale；T_iIt is that estimation on i-th of direction is made an uproar Sound, W_i(x, y) is the weighting function on i-th of direction, f_p(u)=u, u > 0；

Δ φ (x, y)=cos [φ_r(x,y)-φ(x,y)]-|sin[φ_r(x,y)-φ(x,y)]| (2)

φ (x, y) is the local phase weighted mean at the position (x, y), φ_r(x, y) represents local phase at the position (x, y) R grade Fourier expansion series component；

(5) it sums to all gradient magnitudes in G, as final focusing function evaluation of estimate, gradient magnitude summation is higher, currently Picture position is closer to focus position.