CN1598866A - Fish disease automatic diagnosing method processed by image and its device - Google Patents

Abstract

The invention relates to computer digital image process technology in aquiculture industry. The device is made up with printer, upper computer on the spot, image collection card, multi-channel video selector and multiple pick-up heads. Use multi-channel video selector to select some pick-up to shoot rearing pond's real-time picture; the picture is sent to video images collection card; the card converts analog image signal to digital image signal that can be processed by computer; process, identification and alert is made by computer according to preset programs; the processed result is memorized to database in computer; original image is binarized to distinguish white belly of sick fish from background color, then statistic quantity and size of the white zone part, judge is made according to area of white zone to take measures. Adopting the invention, fishes' disease can be early diagnosed and predicted; fishes can be prevented from damage and a new method for aquiculture fishes' disease diagnosis.

Description

Adopt the fish disease automatic diagnosis method and the device thereof of Flame Image Process

Technical field

The present invention relates to the computer digit graph processing technique in aquaculture field, refer in particular to a kind of fish disease automatic diagnosis method and device thereof that adopts Flame Image Process.

Background technology

In recent years, along with expanding economy, the raising day by day of people's living standard, also more and more to the aquatic products demand, so industrialized aquiculture has become the main mode of production of culture fishery.But, a large amount of easily breedings of microorganism in the industrialized aquiculture field, disease is more and more, and the abuse of fish medicine is to the easier accident that causes mass mortality in a short time of fish.Therefore how early detection cultures that pathology or death appear in fish in the pond, so that take its corresponding measures, has become important task in the culture fishery.

Domestic by retrieval Relational database, have many pieces about image processing method application in all fields, as " Shenyang Institute of Technology's journal " 2003.3 " application of Computer Image Processing in diagnostic imaging for medical use ", propose to adopt the real time image data collection, real time image data letter sorting and real time image data treatment technology, pathology flat image outward appearance is carried out multiple processing, and combine with graphic-text database is organic, realization is preserved the data of conditions of patients and is diagnosed; Also as " vibration-testing and diagnosis " 2002.4 " image processing method is in the diesel vibration Application in Fault Diagnosis ", proposition is diagnosed more intuitively for the vibration fault to diesel engine, with diesel engine work circulation is unit, arranging diesel machine surface vibration signal, constitute two dimensional image, utilize the correlation theory of Digital Image Processing, obtain the feature of signal pattern, carry out state recognition and fault diagnosis.

Comprehensive domestic searching document adopts image process method that fish disease is carried out automatic diagnosis and early prediction, does not domesticly see the pertinent literature report, does not more see the scientific and technological achievement report.

Summary of the invention

Purpose of the present invention just provides a kind of fish disease automatic diagnosis method and device thereof, can find in real time to culture that pathology or death appear in fish in the pond, carries out early diagnosis and forecast.

The technical scheme that realizes above-mentioned purpose is:

This device is to be connected to form successively by printer, on-the-spot host computer, image pick-up card, multi-channel video selector switch, a plurality of camera.

This device adopts the fish disease automatic diagnosis method of Flame Image Process:

1. the multi-channel video selector switch is selected certain camera Ti (i=1,2 ..., n), Ti is with fishpond Ci (i=1,2 ..., n) picture be filmed in real time, send video capture card to, image pick-up card is responsible for converting analog picture signal to computing machine treatable data image signal, computing machine is handled, discerns, is reported to the police according to the program that weaves, deposit result in computing machine database, the printer prints picture of publishing picture when needing.

2. will obtain the original image binaryzation, white belly and the background segment of sick fish are opened, add up the number and the size of white portion part then, when the lower limit of minimum white portion more than or equal to the statistical value of this section in the time, and less than going up in limited time of statistical value, promptly thinking has fish uncomfortable reaction or dead to occur, need take measures.

The above-mentioned processing selecting segmentation threshold that carries out the gray level image binaryzation is 225.

The method of the unnecessary white dot that contains in the image after the above-mentioned removal binaryzation is taked this bianry image shrinkage expansion 4 times repeatedly.

The method of above-mentioned expression provincial characteristics is utilized the notion of square in the mechanics, and as particle, the coordinate of pixel is represented with the form of each rank square as the arm of force with the pixel of intra-zone.

The principle of the inventive method is: contaminated when the water quality of fish lives, fish are sick or drug poisoning, occur when uncomfortable or dead, and fish can be turned on one's side, and the belly color can be kept afloat than shallow portion, and the background color darker with the pond has bigger difference, distinguishes easily.And unified fishing gear of putting in a suitable place to breed has kind identical in plant, characteristics such as size is close, this just can utilize statistical method, draws the threshold value of cutting apart disease fish and pond water background colour, also can obtain simultaneously the size of fish in a certain section period pond and the size of belly white portion.

Advantage of the present invention:

The image processing method of computing machine is used in the fish disease automatic diagnosis of aquaculture, can carries out early diagnosis and forecast fish disease.Use this method, can carry out fish disease diagnosis in real time, do not damage and culture the fish body, for the industrialized aquiculture fish disease diagnosis provides new method.

Description of drawings

Fig. 1 is that device of the present invention constitutes synoptic diagram

Fig. 2 image processing system program flow diagram

Image graph behind Fig. 3 original image and the gray processing

Image graph after the binaryzation of Fig. 4 gray level image

Image graph after Fig. 5 Flame Image Process

Embodiment

As shown in Figure 1, this device is to be connected to form successively by a plurality of cameras such as printer 1, on-the-spot host computer 2, image pick-up card 3, multi-channel video selector switch 4, camera T1, T2......, Tn.

According to Fig. 2 flow process, at first computing machine 2 brings into operation, and carries out initialization of variable, and certain camera Ti is taken fishpond Ci real-time pictures, by image pick-up card 3, reads in piece image in computing machine 2 internal memories.Be converted to gray-scale map then,, eliminate noise, carried out medium filtering in order to suppress to disturb.The result who handles as shown in Figure 3.

Then carry out the processing of gray level image binaryzation, the image binaryzation key is to choose segmentation threshold, and sick fish and background segment are opened.Obtain statistics according to repeatedly testing, threshold value is 225 o'clock, can open pending object and background segment preferably, and the image after the binaryzation of gray level image as shown in Figure 4.

Also contain many white dots in the image after the binaryzation, clearly the not sick fish of these white dots is caused, thus must get rid of, otherwise will influence next step processing.Rule of thumb this bianry image can be removed unnecessary white dot behind the shrinkage expansion 4 times repeatedly preferably, the image after the processing as shown in Figure 5.

Then utilize the notion of square in the mechanics again, as particle, the coordinate of pixel is as the arm of force with the pixel of intra-zone, thereby represents provincial characteristics with the form of each rank square.If the quality of each pixel of image is 1, promptly the quality of 1-pixel just equals its pixel value; S is a graphics area; I, j are figure interior pixel coordinate.The formula of square can be expressed as:

$M(p,q)=\underset{(i,j)\∈s}{\mathrm{\Σ}}{i}^p{j}^{q}f(i,j)$

In the formula, M is p, the square of the figure under the q value; P=0,1,2 ...; Q=0,1,2...; (i j) is equivalent to a pixel quality to f, works as p, and q value difference can get the different square of exponent number.If p=0, during q=0:

$M\left(\mathrm{0,0}\right)=\underset{(i,j)\∈s}{\mathrm{\Σ}}f(i,j)$

Be 1-pixel sum, the just area of image in the image.

Carry out the judgement and the warning of fish disease at last,,, can judge the size of 2 white points respectively with 1 and 2 marks with 2 white dots among Fig. 5.According to the experiment statistics data, the white portion area that is exposed when this section wall scroll in period fish disease becomes or be dead is M (0,0)=1000 ± 200, will with Minimum Area in 2 good white portions of 1 and 2 marks respectively with M (0,0) the value lower limit and the upper limit relatively drop between the two as if value, and then the discomfort or the phenomena of mortality appear in fish, system alarm, otherwise normal.

Claims (5)

1. adopt the method for the fish disease automatic diagnosis of image processing method, it is characterized in that:

(1) the multi-channel video selector switch is selected certain camera Ti (i=1,2 ..., n), Ti is with fishpond Ci (i=1,2 ..., n) picture be filmed in real time, send video capture card to, image pick-up card is responsible for converting analog picture signal to computing machine treatable data image signal, computing machine is handled, discerns, is reported to the police according to the program that weaves, deposit result in computing machine database, the printer prints picture of publishing picture when needing;

(2) will obtain the original image binaryzation, white belly and the background segment of sick fish are opened, add up the number and the size of white portion part then, when the lower limit of minimum white portion more than or equal to the statistical value of this section in the time, and less than going up in limited time of statistical value, promptly thinking has fish uncomfortable reaction or dead to occur, carries out early diagnosis and forecast.

2. the method for the fish disease automatic diagnosis of employing image processing method according to claim 1 is characterized in that: the above-mentioned processing selecting segmentation threshold that carries out the gray level image binaryzation is 225.

3. the method for the fish disease automatic diagnosis of employing image processing method according to claim 1 is characterized in that: the method for the unnecessary white dot that contains in the image after the above-mentioned removal binaryzation, take this bianry image shrinkage expansion 4 times repeatedly.

4. the method for the fish disease automatic diagnosis of employing image processing method according to claim 1, it is characterized in that: the method for above-mentioned expression provincial characteristics, utilize the notion of square in the mechanics, with the pixel of intra-zone as particle, the coordinate of pixel is represented with the form of each rank square as the arm of force.

5. realize the device of method of the fish disease automatic diagnosis of the described employing image processing method of claim 1, it is characterized in that: device is connected to form successively by printer, on-the-spot host computer, image pick-up card, multi-channel video selector switch, a plurality of camera.

Images