CN101980297A - Short-distance insect image detection system and detection method thereof - Google Patents

Abstract

The invention discloses a short-distance insect image detection system. The system comprises an image acquisition device and an image processing device, wherein the image acquisition device is used for acquiring a video image of a detection region and transmitting the video image to the image processing device in a wireless mode; and the image processing device is used for processing and analyzing the video image and marking an insect image according to an analysis result. In the short-distance insect image detection system provided by the invention, because the image acquisition device is used for acquiring the video image, and the image processing device is used for processing and analyzing the video image and marking the insect image and the video image which is suspected to be the insect image in real time according to the analysis result, the detection rate of insects is greatly improved, the omission factor of the insects is reduced, meanwhile, the labor intensity of inspection personnel is also reduced.

Description

A kind of closely insect image detection system and detection method thereof

Technical field

The present invention relates to a kind of machine vision technique field, particularly a kind of closely insect image detection system and detection method thereof.

Background technology

In recent years, the harm that causes of biotic intrusion causes widely and payes attention to.According to the data that State Environmental Protection Administration announces, annual China since the direct economic loss that biotic intrusion causes up to 57,400,000,000 yuan, this does not also comprise the ecologic environment loss that causes thus.

In the biology that China has been invaded, the economic loss that insect causes is quite serious.According to conservative estimation, forests such as wet-land pine tree mealybug, Hemiberlesia pitysophila Takagi, fall webworms, Matsumura pine scale invasion insect is seriously taken place, and the annual area that endangers has reached more than 1,500,000 hectares in China; Brontispa longissima dyes worm district area 6,620,000 mu, and the rice water weevil rice area of causing harm has reached 330,000 hectares.Americal rice leaf miner causes economic loss to estimate to reach 4.5 hundred million yuan every year, and the economic loss that colorado potato bug is caused every year is about 1.3 hundred million yuan, and China only to the expenses for prevention and control of one of Americal rice leaf miner, just needs 4.5 hundred million yuan.

Port quarantine is an important ring that prevents the insect invasion, traditional quarantine method mainly adopts visual inspection to search insect, this method is subjected to factor affecting such as environmental baseline, notice scope, visual fatigue, physical activity scope, examination experience, often omission and cause insect to import into.

From statistical data, it is not high that present on-the-spot insect detects the ratio that batch accounts for all quarantines batch, and from the data of above-mentioned insect infringement, actual goods carriage rate should be much larger than these intercepted datas.Above-mentioned insect is made a massive invasion into the fact that causes harm and has proved that omission exists really, and the consequence that causes is very serious.Therefore, strict examination improves recall rate, imports risk into for the reduction insect and is very important.

At present, the approach of strict examination has: 1, increase the time scrutiny, its shortcoming is the time cost height; 2, increase staff, its shortcoming is the cost of labor height; 3, introduce high and new technology, utilize machine vision technique, improve the checking technology means,, in the short period of time, find more insect with less manpower.

But, the algorithm of Machine Vision Recognition insect employing both at home and abroad is fairly simple at present, the needs that can't adapt to port insect quarantine, mainly show as: 1, existing research needs static probe and detects article when detecting, and the place that faces during on-the-spot quarantine is ever-changing, the examination personnel need run in the guide in the middle of the goods, are difficult to find the fixedly place of arrangement for detecting generally speaking; 2, the worm kind of existing research identification is few, requires fixedly insect attitude, and background is single, and on-the-spot quarantine request marks the insect that kind, form, attitude, worm attitude have nothing in common with each other from multiple complex background; 3, on range of application, existing machine vision study limitation is in the identification insect, and the view of the scene requires to increase the mark of the doubtful point of insect, and identifies insect trace complicated and changeable.

And, the equipment that existing research is adopted is too huge, adopt the detecting system that constitutes by video camera and desk-top computer in the research, volume is too big, be not easy to check personnel Deng Che, attend on board, run in the guide in the middle of the goods, and can't detect the insect that is hidden in the place that slit, depths, eminence etc. are not easy to arrive.

As seen, though domestic and international similar research has been pointed out direction for the introduction machine vision technique carries out the insect quarantine, also there is a big difference in the practical application of distance field insect quarantine.

Summary of the invention

In view of above-mentioned the deficiencies in the prior art part, the object of the present invention is to provide a kind of closely insect image detection system, can improve the recall rate of insect.

In order to achieve the above object, the present invention has taked following technical scheme:

A kind of closely insect image detection system wherein, comprising: be used to gather the video image of search coverage, and this video image be transferred to the image collecting device of image processing apparatus by wireless mode; Be used for described video image is handled, analyzed, and mark the image processing apparatus of insect image according to analysis result.

Described closely insect image detection system wherein, also comprises observation eyepiece, and described observation eyepiece is connected with image processing apparatus, is used for the video image that is marked with the insect image of display image treating apparatus output.

Described closely insect image detection system, wherein, image collecting device comprises detecting head, feeler and handle, and an end of described feeler is connected with detecting head, and the other end is connected with handle.

Described closely insect image detection system, wherein, described handle is provided with bluetooth mouse.

Described closely insect image detection system, wherein, described feeler is provided with the support bar that is used to prevent the detecting head shake.

Another object of the present invention also is, the detection method of a kind of closely insect image detection system is provided, and described method may further comprise the steps:

A, gather the video image of search coverage, and this video image is transferred to image processing apparatus wirelessly by image collecting device;

B, described video image is handled, analyzed, and mark the insect image according to analysis result by described image processing apparatus, and the video image in output tape label zone.

Described detection method wherein, also comprises:

C, the video image that is marked with the insect image by observation eyepiece display image treating apparatus output.

Described detection method, wherein, described step B comprises:

B1, employing minimum error method carry out binary conversion treatment to video image, obtain binary image;

B2, search the connected domain of doubtful insect in the binary image;

B3, calculate the complex-shaped degree of described connected domain, and determine the insect image according to the complex-shaped degree of this connected domain.

Described detection method, wherein, described step B1 also comprises:

B11, ask the error function of binary image:

J (t)=P ₀(t) ln σ ₀(t)+P ₁(t) ln σ ₁(t)-P ₀(t) lnP ₀(t)-P ₁(t) lnP ₁(t), wherein

$P_0\left(t\right)=\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right),$ $P_1\left(t\right)=\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right),$

${\mathrm{\σ}}_0\left(t\right)=\frac{\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right){(i-{\mathrm{\μ}}_0\left(t\right))}^2}{P_0\left(t\right)},$ ${\mathrm{\σ}}_1\left(t\right)=\frac{\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right){(i-{\mathrm{\μ}}_1\left(t\right))}^2}{P_1\left(t\right)}$

${\mathrm{\μ}}_0\left(t\right)=\frac{\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right)i}{P_0\left(t\right)},$ ${\mathrm{\μ}}_1\left(t\right)=\frac{\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right)i}{P_1\left(t\right)}$

Wherein, i is a gray-level value, and h (i) is the number of pixels of i for gray shade scale, and t is maximum gray-level value, P ₀(t) in the presentation video gray scale less than the sum of all pixels of t, P ₁(t) represent that gray scale is greater than the sum of all pixels of t, μ in the presentation video ₀And μ ₁The expression gray average, σ ₀And σ ₁Be the gray scale mean square deviation;

B12, will make described error function value maximum gray-level value hour be made as threshold value;

B13, will less than the zone marker of described threshold value the connected domain of doubtful insect.

Described detection method, wherein, the complex-shaped degree of described connected domain is:

$\mathrm{\Ω}=\frac{P}{2*{\mathrm{\π}}^{1/2}*A^{1/2}}$

Wherein, P is the girth of connected domain, and A is the area of connected domain.

Closely insect image detection provided by the invention system, adopted image collecting device to gather video image, image processing apparatus is handled, is analyzed video image, and according to the video image of analysis result real-time mark insect image, doubtful insect, improved the recall rate of insect greatly, reduce the insect loss, reduced examination personnel's labour intensity simultaneously.

Description of drawings

The structured flowchart of the closely insect image detection system that Fig. 1 provides for the embodiment of the invention;

The structural representation of the image collecting device that Fig. 2 provides for the embodiment of the invention;

The structural representation of the detecting head that Fig. 3 provides for the embodiment of the invention;

Fig. 4 is the closely method flow diagram of insect image detection of the present invention.

Embodiment

The invention provides a kind of closely insect image detection system and detection method thereof, this system and method is applicable to that the scene, port carries out container, container car, train, ship and aircraft examination, agricultural product and complementary goods thereof to its loading carry out the insect detection, and are applicable to the investigation of stored grain insects and rice growing season insect.

For making purpose of the present invention, technical scheme and effect clearer, clear and definite, below with reference to accompanying drawing and give an actual example that the present invention is described in more detail.Should be appreciated that specific embodiment described herein only in order to explanation the present invention, and be not used in qualification the present invention.

Whether closely insect image detecting system of the present invention, can detect has insect to exist in the 50cm scope.See also Fig. 1, Fig. 1 is the closely structural representation of insect image detection system of the present invention.It comprises image collecting device 110 and image processing apparatus 120.Described image collecting device 110 and image processing apparatus 120 wireless connections are used to gather the video image of search coverage, and give image processing apparatus with this transmission of video images.Be provided with the video image processing module in the described image processing apparatus 120, be used to receive described video image, and this video image handled, analyzes, draw analysis result, and according to analysis result labeled insect image, for example the image-region with insect in the video image adopts red square frame mark to come out, and shows by display device, makes things convenient for examination personnel examination.

Wherein, described image processing apparatus 120 is a computing machine, its stored the needed data of the view of the scene (comprising known insect image), it can be exported to display with the insect image that marks and show, the examination personnel can find out the position at insect place according to the demonstration of display, and this image processing apparatus can help the examination personnel to overcome the narrow problem of notice scope with a plurality of targets of tense marker in a visual field, has improved examination speed.

Because the detection system that provides of the embodiment of the invention is to survey in-plant insect image, thus moving of need not stopping of its image collecting device, thus change the search coverage of image collecting device.For the ease of control, described image processing apparatus adopts notebook computer, is convenient to carry when the examination personnel detect insect.

Preferably, the closely insect image detection system of the embodiment of the invention also comprises an observation eyepiece 130, this observation eyepiece is the observation eyepiece of spectacle or wear-type, VGA interface (the Video Graphics Array of itself and computing machine, show the drawing array) connect, be used for the video image that is marked with the insect image of real-time display image treating apparatus output.

In other embodiments, described observation eyepiece 130 can also with image processing apparatus 120 wireless connections, like this, the examination personnel are when carrying out the insect detection, only need wear this observation eyepiece 130, come into the examination zone and move described image collecting device 110, obtain the video image of different search coverages.Do not enter the examination scene together and need not carry image processing apparatus 120, and need not be subjected to the length restriction of the connecting line of observation eyepiece.

See also Fig. 2 and Fig. 3, described image collecting device 110 comprises detecting head 111, feeler 112 and handle 113.Described detecting head 111 is provided with camera 1111, illuminating lamp 1112, the brightness regulating switch 1114 that is used to control the camera switch 1113 of camera work and is used to regulate lighting brightness.

Wherein, described camera 1111 is the camera of manual focusing, is mainly used in the video image of gathering in the 10-50cm scope, and in real time to the receiver transmission data that are inserted in the computing machine USB interface, the distance of its transmission can reach 200 meters.For fear of the shielding of wireless transmission, described detecting head 111 and handle 113 all adopt plastic material to make.

Described illuminating lamp 1112 circular array, and be centered around camera 1111 around, the brightness by brightness regulating switch 1114 control illumination can obtain uniform shadowless lighting, improved the sharpness that camera obtains video image.

Wherein, one end of described feeler 112 is connected with detecting head 111 by screw rod 114, detecting head can be rotated on feeler 112, thereby can be according to the angle that detects between needs adjusting detecting head and the feeler, make detecting head 111 test surfaces and goods surface keeping parallelism, and described detecting head 111 is fixing by an adjusting handle 1141.

The other end of described feeler 112 is connected with described handle 113, feeler 112 can reach detecting head 111 in the past examination can't observed place, for example, positions such as the log pile slit of Man Zai container depths, accumulation, freight hold narrow region, eminence goods or goods bottom.And the feeler 112 of present embodiment is elongated, is convenient to pass through in the narrow and small slit.

Because detecting location is apart from difference, so can suitably regulate the length of feeler according to the detection range of reality.The embodiment of the invention provides feeler to comprise rod-pulling type feeler and interconnection system feeler.Wherein, the rod-pulling type feeler also claims telescopic feeler, can directly stretch when regulating length.One end of interconnection system feeler is a screw, and the other end is a nut, headtotail, and its length can infinitely prolong.

Personnel operate image processing apparatus for the ease of examination, described detection system also comprises a bluetooth mouse (not shown), wherein, be provided with a cavity portion (not shown) in the described handle 113, this cavity portion is used to hold described bluetooth mouse and handle 113 is provided with the control button 1131 of controlling described bluetooth mouse.

In the present embodiment, mouse adopts bluetooth approach to be connected with computing machine, and does not adopt wireless mode, mainly is in order to reserve USB interface, to guarantee the speed of camera transmission picture.

When the examination personnel check like this, only need have on observation eyepiece 130, hold feeler 112 and detecting head 111 is placed on tested goods place gets final product.Because in the process of mobile detecting head 111, camera 1111 shakes can cause the image blurring of picked-up, can't carry out image recognition.Present embodiment adopts a support bar 115 is set on feeler 112, can prevent effectively that detecting head 111 from shaking in moving process.

Described support bar 115 can be fixed on the optional position of feeler 112, the adjustable included angle of itself and feeler 112, thus guaranteeing that detecting head 111 and the distance of surveying between object are adjustable, the Support Position is adjustable.And described support bar is provided with roller 1151, and this roller 1151 is a universal wheel, smoothly drags when feeler is surveyed, and realizes supporting continuously.

Proximity detection of the present invention system, be mainly used in and closely detect whether there is insect on the goods, and carry this detecting head and feeler by the examination personnel, come into search coverage and detect, thus require the speed of image processing apparatus processing image fast, the accuracy height, based on above-mentioned closely insect image detection system, the embodiment of the invention is also corresponding to provide the detection method of a kind of closely insect image detection system, and as shown in Figure 4, this method may further comprise the steps:

S100, gather the video image of search coverage, and this video image is transferred to image processing apparatus wirelessly by image collecting device.

Image collecting device adopts the structure of feeler and detecting head combination, can detecting head be placed any search coverage by feeler, and can regulate the brightness of illuminating lamp, thereby obtain video image clearly according to the power of search coverage light.And the present invention adopts Wireless transmission mode, has avoided numerous and diverse cabling, has beautified the outward appearance of image collecting device.

S200, described video image is handled, analyzed, mark the insect image according to analysis result by described image processing apparatus, and the video image in output tape label zone.

This image processing apparatus can be fast carries out mark to doubtful insect in the video image or the zone of confirming as the insect image, for example in the video image that image collecting device obtains, come out with square frame mark red or significantly color in the zone of insect image-region or doubtful insect image, be convenient to the reviewer and find target.

Preferably, this image processing apparatus also can adopt different colors to carry out mark to the zone of insect image-region or doubtful insect image, is convenient to the reviewer different targets is distinguished.

Provide image to compare and the data check function in the described image processing apparatus, by the view of the scene data of inquiry storage and many outstanding experts' examination experience, when operating this image processing apparatus just as there being a plurality of experts to instruct aside, can overcome the examination personnel omits because of the target that forget, ignore, knowledge deficiency etc. causes, reduce the examination capacity variance that the examination personnel produce owing to the equal factor of professional background difference, examination experience, thereby improve the recall rate of insect.Simultaneously, this image processing apparatus can intercept video and save as photo or video file, preserves with the examination recording text, and uploads to long-range server.

S300, the video image that is marked with the insect image by observation eyepiece display image treating apparatus output.

Above-mentioned image processing apparatus generally adopts processing speed computing machine rapidly, and by graphoscope demonstration insect image, because image collecting device is from the close together of detected object, but also need constantly change search coverage, when so the examination personnel check, in the time that search coverage need being come into, and image collecting device need be carried and image processing apparatus is tested, but the display panel of image processing apparatus self is not easy to the examination personnel in the middle use of advancing.Present embodiment adopts the observation eyepiece display image treating apparatus video image that is marked with the insect image of output in real time, the examination personnel mark the zone according to image processing apparatus and check, and this observation eyepiece volume is little, in light weight, be easy to carry, the examination personnel can vacate hand and check.

Wherein, described step S200 specifically comprises:

B 1, employing minimum error method carry out binary conversion treatment to video image, obtain binary image; At first, the error function of definition binary image is:

J (t)=P ₀(t) ln σ ₀(t)+P ₁(t) ln σ ₁(t)-P ₀(t) lnP ₀(t)-P ₁(t) lnP ₁(t), wherein

$P_0\left(t\right)=\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right),$ $P_1\left(t\right)=\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right),$

${\mathrm{\σ}}_0\left(t\right)=\frac{\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right){(i-{\mathrm{\μ}}_0\left(t\right))}^2}{P_0\left(t\right)},$ ${\mathrm{\σ}}_1\left(t\right)=\frac{\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right){(i-{\mathrm{\μ}}_1\left(t\right))}^2}{P_1\left(t\right)}$

${\mathrm{\μ}}_0\left(t\right)=\frac{\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right)i}{P_0\left(t\right)},$ ${\mathrm{\μ}}_1\left(t\right)=\frac{\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right)i}{P_1\left(t\right)}$

Wherein, i is a gray-level value, and h (i) is the number of pixels of i for gray shade scale, and t is maximum gray-level value, its scope between 0-225, P ₀(t) represent that gray scale is less than the sum of all pixels of t, P in the sub-picture ₁(t) expression represents that gray scale is greater than the sum of all pixels of t, μ in the sub-picture ₀And μ ₁The expression gray average, σ ₀And σ ₁Be the gray scale mean square deviation;

Secondly, will make described error function value J (t) maximum gray-level value t hour be made as threshold value; This threshold value is optimal threshold t ^*, that is,

Once more, will less than the zone marker of described threshold value the connected domain of doubtful insect.Because the gray scale of insect is generally less than the gray scale of background, present embodiment will be doubtful zone less than the zone marker of optimal threshold, promptly utilize t ^*As the separatrix, can be with the original image binaryzation.

B2, search the connected domain of doubtful insect in the binary image; After obtaining bianry image, come some interference regions of filtering according to the girth and the area information of each connected domain, search the connected domain of doubtful insect.For example, can be defined as interference region in the zone that connected domain is bigger, it can be filtered out.

B3, calculate the complex-shaped degree of described connected domain, and determine the insect image according to the complex-shaped degree of this connected domain; The complex-shaped degree of described connected domain is:

$\mathrm{\Ω}=\frac{P}{2*{\mathrm{\π}}^{1/2}*A^{1/2}}$

Wherein, P is the girth of connected domain, and A is the area of connected domain.From above-mentioned formula as can be seen, the value of complex-shaped degree Ω is big more, and then the shape of connected domain is complicated more.Because the gray scale of insect is less than the gray scale of background, present embodiment is set the complex-shaped degree of insect connected domain less than 2, therefore the complex-shaped degree of connected domain can be defined as insect less than 2 zone.

In sum, the present invention adopts video image is carried out binary conversion treatment, and binary image is carried out Treatment Analysis, obtains and mark the insect image, and its processing speed is fast, the accuracy height, and realized following beneficial effect:

1, closely insect image detection of the present invention system has adopted elongated feeler and small-sized detecting head, can detect in the past examination can't observed place, as fully loaded container depths, log pile slit, freight hold narrow region, eminence goods or the goods bottom of accumulation, and, this system can be with a plurality of targets of tense marker in a visual field, help the examination personnel to overcome the narrow problem of notice scope, improve examination speed.

2, this closely stable performance of insect image detection system can keep susceptibility for a long time, the visual fatigue problem of similar human eye can not occur, helps to enlarge the examination zone, reduces the dead angle, has increased the probability of finding insect; Also help simultaneously to reduce unloading inspection, save manpower, time and expense.

3, the present invention has adopted machine vision technique that suspicious object all in the visual field is located and mark fast, can help the examination personnel to be primarily focused on rapidly on the suspicious object, accelerates examination speed.

4, insect image detection of the present invention system design is portable, be applicable to step on car, the various examination occasions such as examination and warehouse investigation, field study of attending on board, board.

5, insect image detection of the present invention system loads onto piercing eye for the examination personnel, improve speed and the probability of finding insect, increase recall rate, improve the examination work efficiency, reduced the difference of examination ability between personnel, import risk into thereby reduce insect, accelerate the port watercraft velocity, make plant quarantine view of the scene technology stride forward the machine epoch.

Be understandable that, for those of ordinary skills, can be equal to replacement or change according to technical scheme of the present invention and inventive concept thereof, and all these changes or replacement all should belong to the protection domain of the appended claim of the present invention.

Claims (10)

1. an insect image detection system closely is characterized in that, comprising:

Be used to gather the video image of search coverage, and this video image be transferred to the image collecting device of image processing apparatus by wireless mode;

Be used for described video image is handled, analyzed, and mark the image processing apparatus of insect image according to analysis result.

2. closely insect image detection according to claim 1 system is characterized in that also comprise observation eyepiece, described observation eyepiece is connected with image processing apparatus, is used for the video image that is marked with the insect image of display image treating apparatus output.

3. closely insect image detection according to claim 1 system is characterized in that image collecting device comprises detecting head, feeler and handle, and an end of described feeler is connected with detecting head, and the other end is connected with handle.

4. closely insect image detection according to claim 3 system is characterized in that described handle is provided with bluetooth mouse.

5. closely insect image detection according to claim 3 system is characterized in that, described feeler is provided with the support bar that is used to prevent the detecting head shake.

6. detection method that adopts the described closely insect of claim 1 image detection system to detect the insect image is characterized in that described method may further comprise the steps:

A, gather the video image of search coverage, and this video image is transferred to image processing apparatus wirelessly by image collecting device;

B, described video image is handled, analyzed, mark the insect image according to analysis result by described image processing apparatus, and the video image in output tape label zone.

7. detection method according to claim 6 is characterized in that, also comprises:

C, the video image that is marked with the insect image by observation eyepiece display image treating apparatus output.

8. detection method according to claim 6 is characterized in that, described step B comprises:

B1, employing minimum error method carry out binary conversion treatment to video image, obtain binary image;

B2, search the connected domain of doubtful insect in the binary image;

B3, calculate the complex-shaped degree of described connected domain, and determine the insect image according to the complex-shaped degree of this connected domain.

9. detection method according to claim 8 is characterized in that, described step B1 also comprises:

B11, ask the error function of binary image:

J (t)=P ₀(t) ln σ ₀(t)+P ₁(t) ln σ ₁(t)-P ₀(t) lnP ₀(t)-P ₁(t) lnP ₁(t), wherein

$P_0\left(t\right)=\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right),$ $P_1\left(t\right)=\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right),$

${\mathrm{\σ}}_0\left(t\right)=\frac{\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right){(i-{\mathrm{\μ}}_0\left(t\right))}^2}{P_0\left(t\right)},$ ${\mathrm{\σ}}_1\left(t\right)=\frac{\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right){(i-{\mathrm{\μ}}_1\left(t\right))}^2}{P_1\left(t\right)}$

${\mathrm{\μ}}_0\left(t\right)=\frac{\underset{i=0}{\overset{t}{\mathrm{\Σ}}}h\left(i\right)i}{P_0\left(t\right)},$ ${\mathrm{\μ}}_1\left(t\right)=\frac{\underset{i=t+1}{\overset{255}{\mathrm{\Σ}}}h\left(i\right)i}{P_1\left(t\right)}$

Wherein, i is a gray-level value, and h (i) is the number of pixels of i for gray shade scale, and t is maximum gray-level value, P ₀(t) in the presentation video gray scale less than the sum of all pixels of t, P ₁(t) represent that gray scale is greater than the sum of all pixels of t, μ in the presentation video ₀And μ ₁The expression gray average, σ ₀And σ ₁Be the gray scale mean square deviation;

B12, will make described error function value maximum gray-level value hour be made as threshold value;

B13, will less than the zone marker of described threshold value the connected domain of doubtful insect.

10. detection method according to claim 8 is characterized in that, the complex-shaped degree of described connected domain is:

$\mathrm{\Ω}=\frac{P}{2*{\mathrm{\π}}^{1/2}*A^{1/2}}$

Wherein, P is the girth of connected domain, and A is the area of connected domain.

Images