CN105894532A - Sea surface monitoring image dim target detection method and device - Google Patents

Abstract

An embodiment of the invention discloses a sea surface monitoring image dim target detection method and device which solve a technical problem that conventional four kinds of object detection algorithms which have high requirement for an object and background contrast ratio and an overall image signal to noise ratio cannot be applied to complex background conditions and require a large quantity of cumulative picture frames in quantity accumulation processes and a large amount of computation work.

Description

A kind of sea monitoring image detection method of small target and device

Technical field

The present invention relates to technical field of image processing, particularly relate to a kind of sea monitoring image Weak target inspection Survey method and device.

Background technology

When imaging system distant, imaging plane only accounts for the several or area of ten several pixels, Show as the target that point-like or mottled, contrast and signal to noise ratio are relatively low, referred to as Small object.Fixed by it Justice is it can be seen that " little " in Small object comprises the implication of two aspects of objective attribute target attribute: one is feeling the pulse with the finger-tip Target size is little, and i.e. in image, object pixel quantity is few；Two grey-scale contrast referring to target or signal to noise ratios Low.Target image is separated from background image by target detection exactly, detects in sequence image The Dim targets detection process with relatively low signal-to-noise ratio target should include background suppression, track search and energy Accumulation, Threshold segmentation under the conditions of constant false alarm rate, based on the relevant goal verification etc. of flight path in sequence image Four complete steps.The various Dim targets detection algorithms of current finding report, are the portion of this process Subassembly, the difference is that only that the specific algorithm that each link uses is different, the collocation mode of each link not With.

According to four above-mentioned steps, the common method of target detection is following four kinds.

1, Threshold segmentation

For sea monitoring image, it is considered that the gray scale of the gray scale of target and background and noise is Difference, be the most also simplest object detection method be exactly Threshold segmentation.Set a threshold value, Scene image is considered impact point higher than the pixel of threshold value, and less than the pixel of threshold value be considered background and Noise.Directly carry out that threshold segmentation method is simple, meaning understands, but in actual application and impracticable.By Various background objects is often there is, such as the cloud cluster in sky background, in Sea background in the range of imaging viewing field Wave etc., these background objects tonal gradation in whole scene is often in a higher value model Enclose, even exceed target gray to be detected.Now affect testing result is not only noise, also wraps Including these mixed and disorderly backgrounds, tending not to obtain correct target if directly carrying out Threshold segmentation.

2, background suppression+Threshold segmentation

In view of target shared scope less (one or several pixel) in scene image, it is believed that be high Frequently part；And background objects shows as large area region, it is believed that be low frequency part.Utilize between the two This difference, the background objects that first will appear as low frequency component filters from IR Scene image, only retains Show as target and the noise of radio-frequency component；The most again the image after wiping out background is carried out Threshold segmentation. This image background is filtered and only the process of retention point echo signal and noise be commonly called that " background presses down System ".Due to through background suppress after image signal to noise ratio than not through background suppression image signal to noise ratio High, therefore the Detection results for target can be obviously improved.General by small target deteection at present High-pass filtering method, Order Filtering method, medium filtering, method of least square is had all over the background suppression method used Filtering, matched filtering, background forecast, two dimension Minimum Mean Square Error (LMS) adaptive space predictive filtering, Local entropy method, neural network, wavelet transformation, fractal technology, Mathematical Morphology Method etc..By entering Though the target detection effect that the suppression of row background carries out Threshold segmentation again has certain improvement, but to ensure detection Target detected on the premise of rate and false alarm rate, the signal to noise ratio of image also be there are certain requirements.Detect There is the echo signal of less signal to noise ratio, it is necessary to use lower segmentation threshold that image is split. And along with the reduction of threshold value, more noise spot will be caused to be divided into " target ", make system false alarm rate Increase.

3, the Threshold segmentation under the conditions of background suppression+constant false alarm rate+sequence image detection

The track of target time, there is seriality on sky, show in sequence image to be exactly that impact point is front Position in rear consecutive frame image has dependency.In contrast, the amplitude of noise spot and locus are all Being random, so " decoy " that be partitioned in single-frame images by mistake, it is in consecutive frame image Position will not possess dependency spatially.During therefore available targets and noise spot this, empty dependency Difference rejects " decoy " that amplitude noise is caused, and reduces system false alarm rate.If pursuing inspection Survey the lower target of signal to noise ratio and segmentation threshold is obtained too low, noise can be caused to be divided into " decoy " The quantity of point sharply increase, show as the density of " impact point " in piece image bigger.So, divide The dependency cutting rear target location will be flooded by " decoy point ", rely on track dependency and differentiate true and false Target is by ineffective.Additionally, rear end is for carrying out differentiating the place of target location dependency in real system The disposal ability of reason machine is limited, it is necessary to overall " target " number after segmentation is carried out certain restriction, Both ensured that " target " density after segmentation was unlikely to excessive, ensured again that datatron can process in real time. To this end, " constant false alarm rate " (CFAR) detection method during Radar Signal Processing is incorporated into sequence by people During the detection of row image object, complete the adaptivenon-uniform sampling for image.Use " background suppression+permanent The detection of Threshold segmentation under the conditions of false alarm rate+sequence image " detection method, according to track relevant treatment Algorithm and the difference of processor performance, typically require target signal to noise ratio 3.5～4.5, and segmentation threshold is little In 3.5 time, will result in the decoy density in scene excessive, cause track correlation procedure to lose efficacy.

4, the Threshold segmentation+sequence under the conditions of background suppression+track search and energy accumulation+constant false alarm rate Image detects

Utilize the location-related characteristic of impact point in sequence image to detect, segmentation threshold can be made to reduce, The weak signal target relatively low signal to noise ratio to be detected.But, due to " decoy " density and the relevant place of track The restriction of reason machine disposal ability, determines the threshold signal-to-noise ratio during detection and is subject to certain restrictions, no Reducing to possibility " unrestrictedly ", this most just determines the target minimum noise that detecting system can be detected by Ratio, the size of this value and system noise levels have close relationship.Considering can divided threshold value Signal to noise ratio can not reduce " unrestrictedly ", and cumulative by signal is to improve the maximally effective way of Signal-to-Noise Footpath.Due to the dependency of signal, after n times are cumulative, signal amplitude will increase to original N times；And Noise has randomness, and noise the most in the same time does not the most have dependency, therefore carries out n times and add up After, signal power will increase N times, i.e. amplitude increases N times；That is, after being added up by n times, it is The signal to noise ratio of system will increase N times.Theoretically, as long as cumulative times N is sufficiently large, then low signal-to-noise ratio Echo signal can also be by cumulative and make its signal to noise ratio improve and then exceed detectable minimum signal to noise ratio Threshold value, but this can be limited by many factors in real process.Improve by signal accumulation method Echo signal the most in the same time is accurately added up by primarily premise is that of Signal-to-Noise, as it cannot be guaranteed that This point, makes existing echo signal in accumulated value the most in the same time, has again noise signal, and cumulative process will Lose meaning.Use " the threshold value under the conditions of background suppression+track search and energy accumulation+constant false alarm rate Segmentation+sequence image detects " the detection performance of method, it is decided by that detection algorithm within a certain period of time can Search the frame number N being effective to signal accumulation.Current all kinds of detection algorithm, owing to carrying out signal accumulation During all have the addition of non-targeted signal frame, its performance is far from reaching in theory signal to noise ratio to be improved N Effect again, therefore the raising for detection performance is limited in one's ability.

In sum, above-mentioned four kinds of algorithm of target detection all also exist the limitation of the scope of application, main table Present: the contrast of target and background is required higher by (1)；(2) the overall signal to noise ratio requirement to image Higher；(3) it is not applied for complex background condition；(4) require during energy accumulation to participate in cumulative image Frame number is more；(5) operand is bigger.

Summary of the invention

Embodiments provide a kind of sea monitoring image detection method of small target and device, solve The contrast to target and background that current four kinds of algorithm of target detection are caused requires higher, to image Overall signal to noise ratio require higher, be not applied for when complex background condition, amount accumulation requiring participating in cumulative The technical problem that number of image frames is more, operand is bigger.

A kind of sea monitoring image detection method of small target that the embodiment of the present invention provides, including:

Determine the sea horizon vertical coordinate Y of sea monitoring image, and determine coordinate according to preset original image Initial point O；

Vertical coordinate centered by described sea horizon vertical coordinate Y, sets detection square according to described zero O The detection rectangular dimension of shape, the width of described detection rectangular dimension is W pixel, a length of H pixel；

The described detection rectangle adjacent to the direction each two of the original image abscissa of described original image Gray value summation in the range of described detection rectangular dimension asks for gray scale summation difference, in described original image Including k described detection rectangle；

Determine that described detection rectangle corresponding to maximum gray scale summation difference is according to described gray scale summation difference Rear string detection rectangle abscissa is target lateral coordinates to be measured, and determines that described sea horizon vertical coordinate Y is for treating Survey target vertical coordinate.

Optionally it is determined that the sea horizon vertical coordinate Y of sea monitoring image, and according to preset original image Determine that zero O specifically includes:

By being defined as zero O by the upper left corner of the described original image of sensor acquisition, described former The width of the original image size of beginning image is W₀Individual pixel, total height is H₀Individual pixel；

Determine the described sea horizon vertical coordinate Y of sea monitoring image.

Alternatively, adjacent to the direction each two of the original image abscissa of described original image described inspection Gray value summation in the range of the described detection rectangular dimension of survey rectangle is asked for gray scale summation difference and is specifically wrapped Include:

With described W₀=1 is initial described original image abscissa, along described original image abscissa side To successively according to formulaCalculate each described detection rectangular dimension model Enclosing interior gray value summation, (i j) is the gray value of the i-th row jth row pixel to described P；

It is total that the described gray value summation of the described detection rectangle adjacent to each two asks for k-1 described gray scale With difference DELTA S_1,2,…,k-1。

Alternatively, determine, according to described gray scale summation difference, the described detection that maximum gray scale summation difference is corresponding The last string detection rectangle abscissa of rectangle is target lateral coordinates to be measured, and determines the vertical seat of described sea horizon Mark Y is that target vertical coordinate to be measured specifically includes:

Judge whether described maximum gray scale summation difference exists two equal described maximum gray scale summation differences, If, it is determined that two described detection rectangles that two equal described maximum gray scale summation differences are corresponding, and Center abscissa according to described original imageDetermine the described last string of detection rectangle that distance is the most close Detection rectangle abscissa is target lateral coordinates to be measured；

Determine that described sea horizon vertical coordinate Y is target vertical coordinate to be measured.

Alternatively, it is judged that it is total whether described maximum gray scale summation difference exists two equal described maximum gray scales When being no with difference, then according to described gray scale summation difference determine maximum gray scale summation difference corresponding described in The last string detection rectangle abscissa of detection rectangle is target lateral coordinates to be measured, and determines described sea horizon Vertical coordinate Y is target vertical coordinate to be measured.

Alternatively, described W is 5 pixels, and described H is 41 pixels.

A kind of sea monitoring image Dim targets detection device that the embodiment of the present invention provides, including:

First determines unit, for determining the sea horizon vertical coordinate Y of sea monitoring image, and according to preset Original image determine zero O；

Setup unit, for vertical coordinate centered by described sea horizon vertical coordinate Y, former according to described coordinate Point O sets the detection rectangular dimension of detection rectangle, and the width of described detection rectangular dimension is W pixel, A length of H pixel；

Computing unit is adjacent for the direction each two of the original image abscissa to described original image Gray value summation in the range of the described detection rectangular dimension of described detection rectangle asks for gray scale summation difference, Described original image includes k described detection rectangle；

Second determines unit, for determining that maximum gray scale summation difference is corresponding according to described gray scale summation difference The last string detection rectangle abscissa of described detection rectangle be target lateral coordinates to be measured, and determine described Sea horizon vertical coordinate Y is target vertical coordinate to be measured.

Alternatively, first determines that unit specifically includes:

First determines subelement, for being determined by the upper left corner of the described original image of sensor acquisition For zero O, the width of the original image size of described original image is W₀Individual pixel, total height is H₀Individual pixel；

Second determines subelement, for determining the described sea horizon vertical coordinate Y of sea monitoring image.

Alternatively, computing unit specifically includes:

First computation subunit, for described W₀=1 is initial described original image abscissa, along Described original image abscissa direction is successively according to formulaCalculate each Gray value summation in the range of individual described detection rectangular dimension, (i j) is the ash of the i-th row jth row pixel to described P Angle value；

Second computation subunit, for the described gray value summation of the described detection rectangle adjacent to each two Ask for k-1 described gray scale summation difference DELTA S_1,2,…,k-1。

Alternatively, second determines that unit specifically includes:

3rd determines subelement, is used for judging whether described maximum gray scale summation difference exists two equal institutes State maximum gray scale summation difference, if, it is determined that two equal described maximum gray scale summation differences are corresponding Two described detection rectangles, and according to the center abscissa of described original imageDetermine that distance is the most close Described detection rectangle last string detection rectangle abscissa is target lateral coordinates to be measured, if it is not, then according to institute State gray scale summation difference and determine the last string detection of described detection rectangle corresponding to maximum gray scale summation difference Rectangle abscissa is target lateral coordinates to be measured；

4th determines subelement, is used for determining that described sea horizon vertical coordinate Y is target vertical coordinate to be measured.

As can be seen from the above technical solutions, the embodiment of the present invention has the advantage that

The sea monitoring image that the embodiment of the present invention is absorbed for boat-carrying imaging system, to naval vessel therein mesh Mark detects, and determines target position on the image plane, i.e. determines abscissa and the vertical coordinate of target. According to the feature of boat-carrying imaging system, in sea monitoring image, Ship Target is positioned at sea horizon position, The premise carrying out target detection is to obtain the position of sea horizon, i.e. the vertical coordinate of sea horizon, and this is also we The initial conditions of method.Therefore, in the case of known to sea horizon, the vertical coordinate of Ship Target is sea prison Sea horizon vertical coordinate in altimetric image, Ship Target Detection just determines that the abscissa of Ship Target.This The core concept of bright embodiment is: centered by sea horizon vertical coordinate, open up one occupy one fixed width and The rectangular area of height, and the adjacent of same widths and height is asked successively along the abscissa direction of original image The difference of the pixel sum of rectangular area, center, the rectangular area abscissa of difference maximum is the horizontal stroke of Ship Target Coordinate.The embodiment of the present invention does not carries out any conversion to image, and calculating is only carried out at regional area, meter Calculation amount is little, and accuracy is high.Thus solve that current four kinds of algorithm of target detection are caused to target and the back of the body The contrast of scape require overall signal to noise ratio higher, to image require higher, be not applied for complex background Require to participate in the cumulative technical problem that number of image frames is more, operand is bigger when condition, amount accumulation.

Accompanying drawing explanation

In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to reality Execute the required accompanying drawing used in example or description of the prior art to be briefly described, it should be apparent that below, Accompanying drawing in description is only some embodiments of the present invention, for those of ordinary skill in the art, On the premise of not paying creative work, it is also possible to obtain other accompanying drawing according to these accompanying drawings.

Fig. 1 monitors the one of image detection method of small target for a kind of sea provided in the embodiment of the present invention The schematic flow sheet of individual embodiment；

Fig. 2 monitors the another of image detection method of small target for a kind of sea provided in the embodiment of the present invention The schematic flow sheet of one embodiment；

Fig. 3 monitors the one of image Dim targets detection device for a kind of sea provided in the embodiment of the present invention The structural representation of individual embodiment；

Fig. 4 monitors the another of image Dim targets detection device for a kind of sea provided in the embodiment of the present invention The structural representation of one embodiment.

Detailed description of the invention

Embodiments provide a kind of sea monitoring image detection method of small target and device, solve The contrast to target and background that current four kinds of algorithm of target detection are caused requires higher, to image Overall signal to noise ratio require higher, be not applied for when complex background condition, amount accumulation requiring participating in cumulative The technical problem that number of image frames is more, operand is bigger.

For making the goal of the invention of the present invention, feature, the advantage can be the most obvious and understandable, below will In conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Ground describes, it is clear that the embodiments described below are only a part of embodiment of the present invention, and not all Embodiment.Based on the embodiment in the present invention, those of ordinary skill in the art are not making creativeness The all other embodiments obtained under work premise, broadly fall into the scope of protection of the invention.

Refer to Fig. 1, a kind of sea monitoring image detection method of small target that embodiment of the present invention kind provides An embodiment include:

101, determine the sea horizon vertical coordinate Y of sea monitoring image, and determine according to preset original image Zero O；

102, vertical coordinate centered by sea horizon vertical coordinate Y, sets detection rectangle according to zero O Detection rectangular dimension, detection rectangular dimension width be W pixel, a length of H pixel；

103, the detection of the detection rectangle adjacent to the direction each two of the original image abscissa of original image Gray value summation in the range of rectangular dimension asks for gray scale summation difference, and original image includes k detection Rectangle；

104, last of detection rectangle corresponding to maximum gray scale summation difference is determined according to gray scale summation difference Row detection rectangle abscissa is target lateral coordinates to be measured, and determines that sea horizon vertical coordinate Y is that target to be measured is indulged Coordinate.

The embodiment of the present invention, by centered by sea horizon vertical coordinate, is opened up one and is occupied one fixed width and height The rectangular area of degree, and same widths and the adjacent square of height is asked successively along the abscissa direction of original image The difference of the pixel sum in shape region, center, the rectangular area abscissa of difference maximum is the horizontal seat of Ship Target Mark.The embodiment of the present invention does not carries out any conversion to image, and calculating is only carried out at regional area, calculates Measuring little, accuracy is high.Thus solve that current four kinds of algorithm of target detection are caused to target and background Contrast require higher, the overall signal to noise ratio of image is required higher, be not applied for complex background bar Require to participate in the cumulative technical problem that number of image frames is more, operand is bigger when part, amount accumulation.

The above is that the process to sea monitoring image detection method of small target is described in detail, below Detection rectangle chi by detection rectangle adjacent for the direction each two of the original image abscissa to original image Gray value summation in the range of very little is asked for the calculating process of gray scale summation difference and is described in detail, and please join Readding Fig. 2, another of image detection method of small target is monitored on a kind of sea provided in the embodiment of the present invention Embodiment includes:

201, zero O will be defined as by the upper left corner of the original image of sensor acquisition；

In the present embodiment, when needing to detect the Ship Target that target to be measured is small and weak, needing will be logical The upper left corner of the original image crossing sensor acquisition is defined as zero O, the original image of original image The width of size is W₀Individual pixel, total height is H₀Individual pixel.

202, the sea horizon vertical coordinate Y of sea monitoring image is determined；

203, vertical coordinate centered by sea horizon vertical coordinate Y, sets detection rectangle according to zero O Detection rectangular dimension, detection rectangular dimension width be W pixel, a length of H pixel；

Above-mentioned W is 5 pixels, and H is 41 pixels, and i.e. this detection rectangular area is about sea horizon Symmetrical above and below, the center vertical coordinate of this rectangular area is Y.

205, the gray value summation of the detection rectangle adjacent to each two asks for k-1 gray scale summation difference ΔS_1,2,…,k-1；

Above-mentioned original image includes k detection rectangle.

206, judge whether maximum gray scale summation difference exists two equal maximum gray scale summation differences, if so, Then perform step 207, if it is not, then perform step 208；

207, two detection rectangles that two equal maximum gray scale summation differences are corresponding are determined, and according to original The center abscissa of imageDetermine that the detection rectangle last string detection rectangle abscissa that distance is the most close is Target lateral coordinates to be measured；

208, last of detection rectangle corresponding to maximum gray scale summation difference is determined according to gray scale summation difference Row detection rectangle abscissa is target lateral coordinates to be measured；

209, determine that sea horizon vertical coordinate Y is target vertical coordinate to be measured.

With a concrete application scenarios, the embodiment shown in Fig. 2 is described in detail below, application examples bag Include:

The unit of the length and width below related to is all pixel count.

The vertical coordinate of Ship Target is exactly the sea horizon vertical coordinate in sea monitoring image, at this with variable Y Represent.

With the upper left corner of the original image of sensor acquisition for zero O, line direction is from left to right, Column direction is from the top down, if the original image overall width gathered is W₀, total height is H₀。

Realize step as follows:

Centered by sea horizon, opening up a detection rectangular area, width and height following formula determine:

$\left\{\begin{array}{c}W=5\\ H=41\end{array}\right.$

That is: this detection rectangular area is symmetrical above and below about sea horizon, and the center of this detection rectangular area is vertical to be sat It is designated as Y.

From the first row of original image, calculate the summation of the grey scale pixel value of above-mentioned detection rectangular area, The gray value sum of the 1st rectangular area is

Wherein (i j) is the gray value of the i-th row jth row pixel to P.

Line direction along original image ask successively same widths and height adjacent rectangle region pixel and, The gray value sum in the 2nd adjacent rectangle region isTherefore, the adjacent square of kth The gray value sum in shape region is

Such as, last adjacent detection rectangular area, namelyThe ash of individual detection rectangular area Angle value sum is $S_{\[\frac{W_0}{5}\]}=\underset{i=5\{\[\frac{W_0}{5}\]-1\}+1}{\overset{5\{\[\frac{W_0}{5}\]-1\}+5}{\Σ}}\underset{j=Y-20}{\overset{Y+20}{\Σ}}P(i,j);$

The present embodiment is rounding operation, the most identical, if last detection rectangular area is less than 5 Row, then abandon this rectangular area.

Ask the difference of the pixel sum of adjacent two detection rectangular areas, the 2nd detection rectangular area pixel successively And with the 1st detection rectangular area pixel and difference be Δ S₁=S₂-S₁；

Kth rectangular area pixel and with-1 rectangular area pixel of kth and difference be Δ S_k-1=S_k-S_k-1；

It is thus possible, for instance theIndividual rectangular area pixel and withIndividual rectangular area pixel and it Difference is ${\mathrm{\ΔS}}_{\[\frac{W_0}{5}\]-1}=S_{\[\frac{W_0}{5}\]}-S_{\[\frac{W_0}{5}\]-1};$

WillIt is ranked up according to size, by rectangular area corresponding for maximum The abscissa value of rear string is labeled as X, i.e. X=5k, works as S_max=Δ S_k；

X value is the abscissa of Ship Target.

If equal situation there is, i.e. Δ S_m=Δ S_n；

At this moment, use the method compared with the distance of original image center vertical coordinate, be shown below:

$X=\left\{\begin{array}{cc}5m& |5m-\frac{W_0}{2}|\≤|5n-\frac{W_0}{2}|\\ 5n& |5m-\frac{W_0}{2}|>|5n-\frac{W_0}{2}|\end{array}\right.,$ I.e. select with original image center abscissa apart near rectangle Region is Ship Target region.

And the vertical coordinate of Ship Target is sea horizon vertical coordinate, target detection to be measured is complete.

The embodiment of the present invention, by centered by sea horizon vertical coordinate, is opened up one and is occupied one fixed width and height The rectangular area of degree, and same widths and the adjacent square of height is asked successively along the abscissa direction of original image The difference of the pixel sum in shape region, center, the rectangular area abscissa of difference maximum is the horizontal seat of Ship Target Mark.The embodiment of the present invention does not carries out any conversion to image, and calculating is only carried out at regional area, calculates Measuring little, accuracy is high.Thus solve that current four kinds of algorithm of target detection are caused to target and background Contrast require higher, the overall signal to noise ratio of image is required higher, be not applied for complex background bar Require to participate in the cumulative technical problem that number of image frames is more, operand is bigger when part, amount accumulation.

Refer to Fig. 3, a kind of sea monitoring image Dim targets detection device provided in the embodiment of the present invention An embodiment include:

First determines unit 301, for determining the sea horizon vertical coordinate Y of sea monitoring image, and according in advance The original image put determines zero O；

Setup unit 302, for vertical coordinate centered by sea horizon vertical coordinate Y, according to zero O Setting the detection rectangular dimension of detection rectangle, the width of detection rectangular dimension is W pixel, a length of H Individual pixel；

Computing unit 303, is used for the inspection that the direction each two of the original image abscissa to original image is adjacent Survey the gray value summation in the range of the detection rectangular dimension of rectangle and ask for gray scale summation difference, in original image Including k detection rectangle；

Second determines unit 304, for determining that maximum gray scale summation difference is corresponding according to gray scale summation difference The last string detection rectangle abscissa of detection rectangle is target lateral coordinates to be measured, and determines the vertical seat of sea horizon Mark Y is target vertical coordinate to be measured.

The above is that each unit to sea monitoring image Dim targets detection device is described in detail, under Sub-unit is described in detail by face, refers to Fig. 4, a kind of sea provided in the embodiment of the present invention Another embodiment of monitoring image Dim targets detection device includes:

First determines unit 401, for determining the sea horizon vertical coordinate Y of sea monitoring image, and according in advance The original image put determines zero O；

Further, first determines that unit 401 specifically includes:

First determines subelement 4011, for being determined by the upper left corner of the original image of sensor acquisition For zero O, the width of the original image size of original image is W₀Individual pixel, total height is H₀ Individual pixel；

Second determines subelement 4012, for determining the sea horizon vertical coordinate Y of sea monitoring image.

Setup unit 402, for vertical coordinate centered by sea horizon vertical coordinate Y, according to zero O Setting the detection rectangular dimension of detection rectangle, the width of detection rectangular dimension is W pixel, a length of H Individual pixel；

Computing unit 403, is used for the inspection that the direction each two of the original image abscissa to original image is adjacent Survey the gray value summation in the range of the detection rectangular dimension of rectangle and ask for gray scale summation difference, in original image Including k detection rectangle；

Further, computing unit 403 specifically includes:

First computation subunit 4031, for W₀=1 is initial original image abscissa, along former Beginning image abscissa direction is successively according to formula $S_{1,2,...,k}=\underset{i=W\{\[k\]-1\}+1}{\overset{W(k-1)+W}{\Σ}}\underset{j=Y-\frac{H}{2}}{\overset{Y+\frac{H}{2}}{\Σ}}P(i,j)$ Calculate each detection Gray value summation in the range of rectangular dimension, (i j) is the gray value of the i-th row jth row pixel to P；

Second computation subunit 4032, the gray value summation for the detection rectangle adjacent to each two is asked for K-1 gray scale summation difference DELTA S_1,2,…,k-1。

Second determines unit 404, for determining that maximum gray scale summation difference is corresponding according to gray scale summation difference The last string detection rectangle abscissa of detection rectangle is target lateral coordinates to be measured, and determines the vertical seat of sea horizon Mark Y is target vertical coordinate to be measured.

Further, second determines that unit 404 specifically includes:

3rd determines subelement 4041, be used for judging maximum gray scale summation difference whether exist two equal High-gray level summation difference, if, it is determined that two detections that two equal maximum gray scale summation differences are corresponding Rectangle, and according to the center abscissa of original imageDetermine the last string of detection rectangle that distance is the most close Detection rectangle abscissa is target lateral coordinates to be measured, if it is not, then determine maximum ash according to gray scale summation difference The last string detection rectangle abscissa of the detection rectangle that degree summation difference is corresponding is target lateral coordinates to be measured；

4th determines subelement 4042, is used for determining that sea horizon vertical coordinate Y is target vertical coordinate to be measured.

Should be noted that, the sea monitoring image Dim targets detection in Fig. 3 and embodiment illustrated in fig. 4 Device, may each be such as computer system.

Those skilled in the art is it can be understood that arrive, and for convenience and simplicity of description, above-mentioned retouches The specific works process of the system stated, device and unit, is referred to the correspondence in preceding method embodiment Process, does not repeats them here.

In several embodiments provided herein, it should be understood that disclosed system, device and Method, can realize by another way.Such as, device embodiment described above is only shown Meaning property, such as, the division of described unit, be only a kind of logic function and divide, actual can when realizing There to be other dividing mode, the most multiple unit or assembly can in conjunction with or be desirably integrated into another System, or some features can ignore, or do not perform.Another point, shown or discussed each other Coupling direct-coupling or communication connection can be the INDIRECT COUPLING by some interfaces, device or unit Or communication connection, can be electrical, machinery or other form.

The described unit illustrated as separating component can be or may not be physically separate, makees The parts shown for unit can be or may not be physical location, i.e. may be located at a place, Or can also be distributed on multiple NE.Can select according to the actual needs part therein or The whole unit of person realizes the purpose of the present embodiment scheme.

It addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, Can also be that unit is individually physically present, it is also possible to two or more unit are integrated in a list In unit.Above-mentioned integrated unit both can realize to use the form of hardware, it would however also be possible to employ software function list The form of unit realizes.

If described integrated unit realizes and as independent production marketing using the form of SFU software functional unit Or when using, can be stored in a computer read/write memory medium.Based on such understanding, this The part that the most in other words prior art contributed of technical scheme of invention or this technical scheme Completely or partially can embody with the form of software product, this computer software product is stored in one In storage medium, including some instructions with so that computer equipment (can be personal computer, Server, or the network equipment etc.) perform completely or partially walking of method described in each embodiment of the present invention Suddenly.And aforesaid storage medium includes: USB flash disk, portable hard drive, read only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disc or CD Etc. the various media that can store program code.

The above, above example only in order to technical scheme to be described, is not intended to limit； Although being described in detail the present invention with reference to previous embodiment, those of ordinary skill in the art should Work as understanding: the technical scheme described in foregoing embodiments still can be modified by it, or to it Middle part technical characteristic carries out equivalent；And these amendments or replacement, do not make appropriate technical solution Essence depart from various embodiments of the present invention technical scheme spirit and scope.

Claims (10)

1. a sea monitoring image detection method of small target, it is characterised in that including:

Determine the sea horizon vertical coordinate Y of sea monitoring image, and determine coordinate according to preset original image Initial point O；

Vertical coordinate centered by described sea horizon vertical coordinate Y, sets detection square according to described zero O The detection rectangular dimension of shape, the width of described detection rectangular dimension is W pixel, a length of H pixel；

The described detection rectangle adjacent to the direction each two of the original image abscissa of described original image Gray value summation in the range of described detection rectangular dimension asks for gray scale summation difference, in described original image Including k described detection rectangle；

Determine that described detection rectangle corresponding to maximum gray scale summation difference is according to described gray scale summation difference Rear string detection rectangle abscissa is target lateral coordinates to be measured, and determines that described sea horizon vertical coordinate Y is for treating Survey target vertical coordinate.

Sea the most according to claim 1 monitoring image detection method of small target, it is characterised in that Determine the sea horizon vertical coordinate Y of sea monitoring image, and determine zero O according to preset original image Specifically include:

By being defined as zero O by the upper left corner of the described original image of sensor acquisition, described former The width of the original image size of beginning image is W₀Individual pixel, total height is H₀Individual pixel；

Determine the described sea horizon vertical coordinate Y of sea monitoring image.

Sea the most according to claim 2 monitoring image detection method of small target, it is characterised in that Described in the described detection rectangle adjacent to the direction each two of the original image abscissa of described original image Gray value summation in the range of detection rectangular dimension is asked for gray scale summation difference and is specifically included:

With described W₀=1 is initial described original image abscissa, along described original image abscissa side To successively according to formulaCalculate each described detection rectangular dimension model Enclosing interior gray value summation, (i j) is the gray value of the i-th row jth row pixel to described P；

It is total that the described gray value summation of the described detection rectangle adjacent to each two asks for k-1 described gray scale With difference DELTA S_1,2,…,k-1。

Sea the most according to claim 3 monitoring image detection method of small target, it is characterised in that Last of described detection rectangle corresponding to maximum gray scale summation difference is determined according to described gray scale summation difference Row detection rectangle abscissa is target lateral coordinates to be measured, and determines that described sea horizon vertical coordinate Y is mesh to be measured Mark vertical coordinate specifically includes:

Judge whether described maximum gray scale summation difference exists two equal described maximum gray scale summation differences, If, it is determined that two described detection rectangles that two equal described maximum gray scale summation differences are corresponding, and Center abscissa according to described original imageDetermine the described last string of detection rectangle that distance is the most close Detection rectangle abscissa is target lateral coordinates to be measured；

Determine that described sea horizon vertical coordinate Y is target vertical coordinate to be measured.

Sea the most according to claim 4 monitoring image detection method of small target, it is characterised in that It is no for judging whether described maximum gray scale summation difference exists two equal described maximum gray scale summation differences Time, then determine described detection rectangle corresponding to maximum gray scale summation difference according to described gray scale summation difference Last string detection rectangle abscissa is target lateral coordinates to be measured.

Monitoring image Dim targets detection side, sea the most as claimed in any of claims 1 to 5 Method, it is characterised in that described W is 5 pixels, described H is 41 pixels.

7. a sea monitoring image Dim targets detection device, it is characterised in that including:

First determines unit, for determining the sea horizon vertical coordinate Y of sea monitoring image, and according to preset Original image determine zero O；

Setup unit, for vertical coordinate centered by described sea horizon vertical coordinate Y, former according to described coordinate Point O sets the detection rectangular dimension of detection rectangle, and the width of described detection rectangular dimension is W pixel, A length of H pixel；

Computing unit is adjacent for the direction each two of the original image abscissa to described original image Gray value summation in the range of the described detection rectangular dimension of described detection rectangle asks for gray scale summation difference, Described original image includes k described detection rectangle；

Second determines unit, for determining that maximum gray scale summation difference is corresponding according to described gray scale summation difference The last string detection rectangle abscissa of described detection rectangle be target lateral coordinates to be measured, and determine described Sea horizon vertical coordinate Y is target vertical coordinate to be measured.

Sea the most according to claim 7 monitoring image Dim targets detection device, it is characterised in that First determines that unit specifically includes:

First determines subelement, for being determined by the upper left corner of the described original image of sensor acquisition For zero O, the width of the original image size of described original image is W₀Individual pixel, total height is H₀Individual pixel；

Second determines subelement, for determining the described sea horizon vertical coordinate Y of sea monitoring image.

Sea the most according to claim 7 monitoring image Dim targets detection device, it is characterised in that Computing unit specifically includes:

First computation subunit, for described W₀=1 is initial described original image abscissa, along Described original image abscissa direction is successively according to formulaCalculate each Gray value summation in the range of individual described detection rectangular dimension, (i j) is the ash of the i-th row jth row pixel to described P Angle value；

Second computation subunit, for the described gray value summation of the described detection rectangle adjacent to each two Ask for k-1 described gray scale summation difference DELTA S_1,2,…,k-1。

Sea the most according to claim 7 monitoring image Dim targets detection device, its feature exists In, second determines that unit specifically includes:

3rd determines subelement, is used for judging whether described maximum gray scale summation difference exists two equal institutes State maximum gray scale summation difference, if, it is determined that two equal described maximum gray scale summation differences are corresponding Two described detection rectangles, and according to the center abscissa of described original imageDetermine that distance is the most close Described detection rectangle last string detection rectangle abscissa is target lateral coordinates to be measured, if it is not, then according to institute State gray scale summation difference and determine the last string detection of described detection rectangle corresponding to maximum gray scale summation difference Rectangle abscissa is target lateral coordinates to be measured；

4th determines subelement, is used for determining that described sea horizon vertical coordinate Y is target vertical coordinate to be measured.