CN101427912B - Drop needle detecting device with handle anastomat punch head and detecting method thereof - Google Patents

Abstract

The invention relates to a drop stitch detecting device on the head of a stapler with a hand lever and a detection method thereof. The drop stitch detecting device comprises a computer, a video camera, an illuminant light and a stapler locating device; the video camera comprises a CMOS picture sensor and a lens; the stapler locating device comprises a pedestal, a stapler clamp, an illuminant rack, a video camera rack, an upright post and a winding gear; the stapler clamp comprises a sliding panel, a component sliding stand, a teeth bar, a fixed pedestal, a dirigible wheel, locking apparatus and a ledger plate; the sliding panel is fixedly arranged on the sliding stand; the ledger plate is fixedly arranged on the fixed pedestal; the teeth bar is arranged on the component sliding stand; andthe dirigible wheel and the locking apparatus are arranged on a lateral surface. Based on the reflection image principle, the drop stitch detecting device adopts a low-angle ring light frontal projection manner for detection. The drop stitch detecting device and the detection method thereof solve the drop stitch detection problem of the stapler with the hand lever, increase the product qualified rate and safety, avoid the human misjudgment, increase the detection efficiency, ensure the product quality, have the advantages of high accuracy, high speed and compact structure and design. Furthermore, the invention is suitable for popularization and use.

Description

The dropped stitch checkout gear and the detection method thereof of band handle anastomat head

Technical field

The present invention relates to a kind of instrument that is used to detect the anastomat dropped stitch, especially a kind of band handle anastomat dropped stitch on-line automatic detection device and detection method thereof based on the catoptric imaging principle.

Background technology

An anastomat shell of pushing pin is that plastic material is made, the seam nail that has two rows of circular arrangement to intermesh in the needle stand head needle tray, and the back connects firing handle.In the operation on digestive tract, hold handle and will stitch the nail percussion, the seam that is driven into organizationally nail, and cut off with the tissue of belt cutter with its inboard simultaneously, can carry out coincideing of luminal organs very easily.Anastomat seam nail is made by Titanium or tantalum, the seam nail very thin (0.11～0.13mm), quantity many (at least 16), nail lower limb weak point very easily come off.Then may cause stitching incomplete if leak nail, complication such as hemorrhage, the narrow and intestinal fistula of anastomotic stoma take place, directly influence therapeutic effect, jeopardize patient's life and health.Therefore, the anastomat dropped stitch detects to become and ensures the anastomat quality, effectively prevents seepage, lowers the key of the incidence rate of anastomotic leakage.Traditional human eye detection wastes time and energy, and can not guarantee to detect effect, forms erroneous judgement under the situation that testing staff's fatigue or energy are not concentrated easily.The external X-ray check that adopts, though the result is accurate, X-radiation can influence the healthy of testing staff and detect cost higher.

Chinese invention patent application number 200510111564.3,2006.06.2 disclose a kind of needle drop detection system for anastomat, and it comprises computer, video camera, illuminator, anastomat positioner, and video camera is made up of cmos image sensor and CMOS photographic head; The anastomat positioner comprises base, clamping fixture seat, the anastomat anchor clamps, lens bracket, guide rail, elevating mechanism, be provided with dovetail column guide rail above the base, living mutually with dovetail guide is equipped with elevating mechanism above the dovetail groove loose piece join, and is provided with lens bracket above the elevating mechanism, is fixedly connected with clamping fixture seat above the base before the guide rail, clamping fixture seat is fixedlyed connected with anchor clamps by dovetail groove, place clouded glass on the anchor clamps lower base, be equipped with light source below the clouded glass, be fixed on the camera lens center on the lens bracket, the anchor clamps center, measured workpiece center in the anchor clamps is on same vertical line.This system can discern the element of dropped stitch effectively, judges the number of dropped stitch, can guarantee the effectiveness of detection from accuracy and accuracy two aspects, and have speed characteristics fast, simple for structure.But this system is not suitable for the dropped stitch of band handle anastomat head to be detected, its technical problem that mainly exists is that this system adopts the back-light lighting system, the transmission imaging principle can't be gathered the original image of band handle anastomat, can not be used to detection, and the anastomat anchor clamps in the anastomat positioner can't satisfy the clamping and the adjusting of band handle anastomat with the handle anastomat.

Summary of the invention

The present invention will provide a kind of dropped stitch checkout gear and detection method thereof with handle anastomat head, these apparatus and method can realize detecting with the dropped stitch of handle anastomat effectively, the element of identification dropped stitch, judge the number of dropped stitch, can guarantee the effectiveness of detection from accuracy and accuracy two aspects, and have speed characteristics fast, simple for structure.

Technical scheme of the present invention is:

A kind of dropped stitch checkout gear with handle anastomat head comprises computer, video camera, illuminator, the anastomat positioner be made up of cmos image sensor and optical lens; The anastomat positioner is by base, the anastomat anchor clamps, light source shelf, camera pedestal, column, elevating mechanism is formed, and video camera is connected by the USB2.0 USB with computer, fixedly connected column above the base, be characterized in: illuminator is based on the annular light source lamp of the low angle face-on projection of catoptric imaging principle, be connected and installed with the camera pedestal of video camera above the column from top to bottom successively by elevating mechanism, the light source shelf of illuminator be installed, the anastomat anchor clamps; Wherein, the measured workpiece center in camera lens center, annular light source lamp center, anastomat anchor clamps center, the anastomat anchor clamps is on same vertical line; The anastomat anchor clamps comprise sliding panel, sliding stand, tooth bar, firm banking, adjustment wheel, locking device, fixing head, fixedly mount sliding panel on the sliding stand, fixedly mount fixing head on the firm banking, the recess of the clamping band handle anastomat head of fixing head and sliding panel is inwardly corresponding, and sliding stand is equipped with tooth bar, and the sliding stand side is equipped with adjusts wheel and locking device.

The highest resolution 2048*1536 of cmos image sensor, Pixel Dimensions are 3.2 μ m*3.2 μ m, and optical dimensions is 1/2inch, and signal to noise ratio＞42dB, dynamic range＞60dB, sensitivity are 1V/lux-sec under the light source of 550nm.

The focal length of optical lens is 55mm, image full-size 8.8mm*6.6mm, aperture F2.8-F32C, focus 0.14m-Inf, operating distance 140mm-Inf, optical magnification 0.5x.

Illuminator is a low angle annular LED light source.

Light source shelf is to be made up of five rectangular strips, wherein, article one, be fixed with connecting rod above the rectangular strip, the two limit symmetries a pair of connecting rod of forming by two rectangular strips that is rotatably connected, be formed on four degree of freedom activities all around, outermost two rectangular strips are provided with the M4 screw, are connected by bolt and annular light source lamp.

A kind of dropped stitch detection method with handle anastomat head is characterized in that it is as follows specifically to detect step:

(5) cmos sensor is gathered original image, and in original image, automatically extract annular related interests zone, the utilization gray scale is cut apart the coarse localization relevant range, utilizes and looks for round operator and radius to carry out match automatically, the annular target area that accurate in locating will detect;

(6) utilize the dual threshold grey level histogram of target area, finish automatic threshold and cut apart, in circle ring area, extract the needle tray region-of-interest, wherein:

Adopt circular expander template in the morphology processing, image can produce a Minkowski increment,

Effectively remove the noise that the needle tray edge reflections is produced, its expression formula is as follows:

$A\⊕B=\∪\{A+b:b\∈B\}---\left(1\right)$

Wherein, A is the high-brightness region that extracts, and B is a structural element, select radius be the circle of R as template, translation in the scope of A, the image after obtaining expanding

Image negate after will expanding just can be removed noise when separating high-brightness region, effectively only contained the region-of-interest that detects the needle tray of being concerned about;

(7) the local threshold partitioning algorithm based on fuzzy set theory is used in image characteristics extraction,

Mean filter smooth template to region-of-interest image employing 15 * 15 strengthens:

$g(m,n)=\frac{1}{15\×15}\mathrm{\Σ\Σf}(m-k,n-l),(k,l)\∈W---\left(2\right)$

F (m, n) and g (m n) is input picture and output image respectively, W be in the position (m, the suitable neighborhood of selecting around pixel n), i.e. pixel in 15 * 15 the spectral window scope, the feature pin that smoothing processing back edge step is bigger, grey scale change is obvious; Comparatively fuzzy base reflection smooth effect relatively a little less than, gray value changes little; Use the soft-threshold of the result of smothing filtering as fuzzy operation again, the degree that belongs to concrete zone according to each pixel is carried out local threshold and is cut apart:

$h(x,y)=\left\{\begin{array}{c}255,f(x,y)\&GreaterEqual;g(x,y)+\mathrm{offset};\\ 0,f(x,y)<g(x,y)+\mathrm{offset};\end{array}\right.---\left(3\right)$

With original image f (x, y) with level and smooth back image g (x, difference y) extracts greater than the image of offset, offset ≠ 0 wherein, choosing of the selection of template and offset has very big motility, can effectively extract the anastomat suture needle of shades of colour model;

(8) adopt area features in the recognition image to judge whether dropped stitch of this needle tray, testing product is divided into i ∈ { I, II| dropped stitch; Pin is arranged } two classes; For the stability of increase differentiating, carry out expansion process earlier, fill the fracture that produces in the cutting procedure, make each feature pin form the connected region of an integral body, the image of each connected region be F (x, y), judge successively according to each connected region area:

$M=\underset{x}{\mathrm{\&Sigma;}}\underset{y}{\mathrm{\&Sigma;}}F(x,y)---\left(4\right)$

M is a cell area, and output suture needle number judges whether product is qualified.

The invention has the beneficial effects as follows:

This device is by real time computer control, and cmos camera is controlled by software-driven, observes the image of anastomat by video camera initialization and real-time display module; To image strengthen, processing such as morphological erosion; At the feature in the pin zone that coincide, information in the local threshold segmented extraction image; Characteristic parameter adopts suitable criterion of identification to detect the dropped stitch situation, utilization Digital Image Processing and feature identification technique, and testing result can intuitively show that also can store easily also can be by the printer output form by display.Because the anastomat anchor clamps adopt sliding panel, sliding stand, tooth bar, firm banking, adjustment wheel, locking device, fixing head, fixedly mount sliding panel on the sliding stand, fixedly mount fixing head on the firm banking, the recess of the clamping band handle anastomat head of fixing head and sliding panel is inwardly corresponding, make and be with handle anastomat clamping and easy to adjust, can satisfy fully the requirement of the dropped stitch detection of being with handle anastomat head, and have simple in structurely, regulate characteristics such as easy to use.

In image resolution ratio is 2048*1536, normal picking rate, choose best conditions of exposure and gain coefficient, white balance adjusting adopts under the condition of the empirical value of testing, 200 anastomats to four kinds of models carry out online detection, dropped stitch discrimination 100%, average time, 2.385s was respond well.

The present invention is a cover accurate detection device and a detection method thereof, and the dropped stitch that can solve band handle anastomat detects problem, improves product percent of pass and safety, avoids artificial erroneous judgement, improves detection efficiency, guarantees the quality of product.Therefore, have that accuracy height, speed are fast, the advantage of decoration structure features simple design, be fit to promote the use of.

Description of drawings

Fig. 1 is a structural upright general assembly sketch map of the present invention;

Fig. 2 anastomat anchor clamps schematic perspective view;

Fig. 3 sliding stand internal structure schematic perspective view;

Fig. 4 lamp source anchor clamps and lamp source assembling sketch map;

Fig. 5 is the workflow diagram of system;

Fig. 6 is the sketch map of the key technology of feature identification;

Fig. 7 is the testing process figure of system.

The specific embodiment

The invention will be further described below in conjunction with accompanying drawing and embodiment.

As shown in Figure 1, the dropped stitch checkout gear based on the band handle anastomat head of catoptric imaging principle comprises computer 1, video camera 2, illuminator 3, anastomat positioner 4, and video camera 2 is made up of cmos image sensor and optical lens.Anastomat positioner 4 comprises base 5, anastomat anchor clamps 6, light source shelf 7, camera pedestal 8, column 9, elevating mechanism 10,11,12.

Fixed installation column 9 above the base 5, elevating mechanism 10,11,12 is installed above the column 9 from top to bottom, be installed with lens bracket 8 above the elevating mechanism 10, be installed with light source shelf 7 above the elevating mechanism 11, stationary annular illuminator 3 on the light source shelf 7 is installed with anastomat anchor clamps 6 above the elevating mechanism 12.

The measured workpiece center in video camera 2 optical centers on the camera pedestal 8, annular light source lamp 3 centers, anastomat anchor clamps 6 centers, the anastomat anchor clamps 6 of being fixed on is on same vertical line; Video camera 2 is connected by the USB2.0 USB with computer 1.

Computer (1) operating system is WindowsXP+SP2, and used software is independently developed multiple digital image processing method and feature identification software.

This device adopts CMOS (complementary metal oxide semiconductors (CMOS), ComplementaryMetal-Oxide Semiconductor) video camera 2 images acquired of sensor devices formation, unification, succinct light path and machinery have been designed voluntarily, use multiple digital image processing method and feature identification technique, realized anastomat dropped stitch ONLINE RECOGNITION.

(1) the conformability height of cmos image sensor and peripheral circuit, volume is less, and power consumption is lower, and transmission speed can be very fast.The high resolution digital camera 2 that adopts has been finished the integration of ADC, capture card, control chip and circuit, volume is little, in light weight, easy to carry, and and the interface of computer 1 adopts the USB2.0 USB, transmission speed is up to 480Mb/s, is fit at a high speed, real-time designing requirement.

Because tested element is had any different according to the different big or small colors of model, circular maximum gauge 34mm, seam nail 0.2～0.3mm for improving the resolution of institute's pickup image, has determined the parameter of optics.

The major parameter of the selected optical element of table 1

(2) low angle annular LED light source, highdensity red LED arrayed is in umbrella-shaped structure, light in lighting source central area high concentration, add the diffuse-reflectance circle on light source, can clearly manifest the anastomat details, pick off is only collected the diffuse-reflectance uniform light simultaneously, reduce the direct reflection of object, increase contrast, the image of avoiding total reflection to cause is bright excessively, and characteristic body is difficult to differentiate.25 ℃ of room temperatures, continue under the situation with 50% brightness illumination, led light source surpasses 30,000 hours service life, and intensity of illumination is stable.

(3) purpose of design of this machinery is mainly photographic head and detected element provides a stable good working environment, so that the accurately collection of location and image.As shown in Figure 1, the measured workpiece center is a work centre, guarantee optical center, light source center, anchor clamps center on same vertical line, working face is the work piece holder upper limb, utilizes elevating mechanism 10,11,12 to realize the flexible of operating distance between video camera, light source, measured workpiece.

The anchor clamps 6 (as Fig. 2) of anastomat by sliding stand 14, tooth bar 15, firm banking 16, adjustment wheel 17, locking device 18, form a cover optical adjusting frame.Guarantee the center of measured workpiece and work centre and the working face that tested surface is whole machinery.Fixed installation sliding panel 13 on the sliding stand 14, fixed installation fixing head 19 on the firm banking 16, sliding stand 14 is equipped with tooth bar 15, and the sliding stand side is equipped with adjusts wheel 17 and locking device 18.Sliding stand internal structure (as Fig. 3) is made of 20 gears, 21 assembling adjustment bars, 22 locking holes.Locking device guarantees to detect the workpiece horizontal stable.

Light source shelf 7 (as Fig. 4) is to be made up of five rectangular strips, and wherein, one is fixed with connecting rod above the rectangular strip, and the two limit symmetries a pair of connecting rod of being made up of two rectangular strips that is rotatably connected forms four degree of freedom activities all around within the specific limits.On outermost two rectangular strips the M4 screw is arranged, available bolt and annular light source lamp are connected.

(4) workflow of whole device (Fig. 5) is as follows: initialization apparatus; Images acquired is to internal memory (in real time and single frames collection, stop, reading, store); Feedback is finished in the real-time parameter setting; Flame Image Process and differentiation; Testing result (storage, printing).

The running environment of systems soft ware is: the microprocessor of 2G internal memory, 160GB hard disk, 2.33GHz dominant frequency, operating system are the WindowsXP+SP2 Chinese edition, adopt Visual C++6.0 making software.This systems soft ware can be provided with and regulate in real time a plurality of parameters (resolution, image window, exposure, gain, white balance, Gamma correction etc.) of photographic head, makes the image acquisition quality reach best; From the digital picture that obtains, extract feature, realize that as criterion high accuracy, high-precision dropped stitch detect perfect in shape and function, reliable, friendly interface, easy to use.

(5) Digital Image Processing and feature identification key technology

The present invention has proposed a kind of method (Fig. 7) that detects automatically based on the anastomat dropped stitch of Flame Image Process and feature identification technique from the angle of computer vision.Concrete algorithm flow is as follows:

(9) in the original image that cmos sensor is gathered, extract annular related interests zone automatically.The utilization gray scale is cut apart the coarse localization relevant range, utilizes and looks for round operator and radius to carry out match automatically, the annular target area that accurate in locating will detect.

(10) utilize the dual threshold grey level histogram of target area, finish automatic threshold and cut apart, in circle ring area, extract the needle tray region-of-interest.

Adopt circular expander template in the morphology processing, image can produce a Minkowski increment,

Effectively remove the noise that the needle tray edge reflections is produced.Its expression formula is as follows:

$A\&CirclePlus;B=\&cup;\{A+b:b\&Element;B\}---\left(1\right)$

Wherein, A is the high-brightness region that extracts, and B is a structural element.Select radius be the circle of R as template, translation in the scope of A, the image after obtaining expanding

Image negate after will expanding just can be removed noise when separating high-brightness region, effectively only contained the region-of-interest that detects the needle tray of being concerned about.

(11) image characteristics extraction is the core of software system, and the feature identification key technology is to propose the local threshold partitioning algorithm (Fig. 6) of utilization based on fuzzy set theory.

Mean filter smooth template to region-of-interest image employing 15 * 15 strengthens:

$g(m,n)=\frac{1}{15\&times;15}\mathrm{\&Sigma;\&Sigma;f}(m-k,n-l),(k,l)\&Element;W---\left(2\right)$

F (m, n) and g (m n) is input picture and output image respectively, and W is (m, the suitable neighborhood of selecting around pixel n), i.e. pixel in 15 * 15 the spectral window scope in the position.The feature pin that smoothing processing back edge step is bigger, grey scale change is obvious; Comparatively fuzzy base reflection smooth effect relatively a little less than, gray value changes little.Use the soft-threshold of the result of smothing filtering as fuzzy operation again, the degree that belongs to concrete zone according to each pixel is carried out local threshold and is cut apart

$h(x,y)=\left\{\begin{array}{c}255,f(x,y)\&GreaterEqual;g(x,y)+\mathrm{offset};\\ 0,f(x,y)<g(x,y)+\mathrm{offset};\end{array}\right.---\left(3\right)$

(x, y) (x, difference y) extracts greater than the image of offset with level and smooth back image g with original image f.Offset ≠ 0 wherein.Choosing of the selection of template and offset has very big motility, can effectively extract the anastomat suture needle of shades of colour model.

(12) adopt the most effective characteristic parameter (detecting the area features in the recognition image)

Judge whether dropped stitch of this needle tray, testing product is divided into i ∈ { I, II| dropped stitch; Pin is arranged } two classes.For the stability of increase differentiating, carry out expansion process earlier, fill the fracture that produces in the cutting procedure, make each feature pin form the connected region of an integral body, the image of each connected region be F (x y), judges successively according to each connected region area,

$M=\underset{x}{\mathrm{\&Sigma;}}\underset{y}{\mathrm{\&Sigma;}}F(x,y)---\left(4\right)$

M is a cell area, and output suture needle number judges whether product is qualified.

The external situation that adopts X-ray check anastomat dropped stitch, although degree of accuracy is very high, but may there be the danger of radiation leakage, operator are brought potential threat, then prolong with artificialvision whether see the leakage detection pin in the process of domestic enterprise's production anastomat always, not only take energy but also time-consuming, and can not guarantee to detect effect.

According to actual working environment, image resolution ratio is 2048*1536, normal picking rate, choose best conditions of exposure and gain coefficient, white balance adjusting adopts under the condition of empirical value of test, and 200 anastomats of 4 kinds of models are carried out online detection, dropped stitch discrimination 100%, average time, 2.385s was respond well.

This device can be discerned the element of dropped stitch effectively, and can judge the number of dropped stitch, can guarantee the effectiveness that detects from accuracy and accuracy two aspects.The noncontact quality testing problem of the band handle anastomat finished product that solves improves production practices efficient, has developed the application of digital image processing techniques, has that accuracy height, speed are fast, the advantage of decoration structure features simple design, is fit to promote the use of.

Claims (6)

1. the dropped stitch checkout gear with handle anastomat head comprises computer (1), video camera (2), illuminator (3), the anastomat positioner (4) be made up of cmos image sensor and optical lens; Described anastomat positioner (4) is by base (5), anastomat anchor clamps (6), light source shelf (7), camera pedestal (8), column (9), elevating mechanism (10,11,12) form, fixedly connected column (9) above the base (5), described video camera (2) is connected by the USB2.0 USB with computer (1), it is characterized in that, described illuminator (3) is based on the annular light source lamp of the low angle face-on projection of catoptric imaging principle, pass through elevating mechanism (10 above the described column (9) from top to bottom successively, 11,12) be connected and installed with the camera pedestal (8) of video camera (2), the light source shelf (7) of annular light source lamp (3) is installed, anastomat anchor clamps (6); Wherein, the measured workpiece center in video camera (2) optical center, annular light source lamp (3) center, anastomat anchor clamps (6) center, the anastomat anchor clamps (6) is on same vertical line; Described anastomat anchor clamps (6) comprise sliding panel (13), sliding stand (14), tooth bar (15), firm banking (16), adjustment wheel (17), locking device (18), fixing head (19), sliding stand (14) is gone up fixed installation sliding panel (13), firm banking (16) is gone up fixed installation fixing head (19), and the recess of the clamping band handle anastomat head of fixing head (19) and sliding panel (13) is inwardly corresponding; Sliding stand (14) is equipped with tooth bar (15), and the sliding stand side is equipped with adjusts wheel (17) and locking device (18).

2. the dropped stitch checkout gear of band handle anastomat head according to claim 1, it is characterized in that, the highest resolution 2048*1536 of described cmos image sensor, Pixel Dimensions is 3.2 μ m*3.2 μ m, optical dimensions is 1/2inch, signal to noise ratio＞42dB, dynamic range＞60dB, sensitivity are 1V/lux-sec under the light source of 550nm.

3. the dropped stitch checkout gear of band handle anastomat head according to claim 1 is characterized in that the focal length of described optical lens is 55mm, image full-size 8.8mm*6.6mm, aperture F2.8-F32C, focus 0.14m-Inf, operating distance 140mm-Inf, optical magnification 0.5x.

4. the dropped stitch checkout gear of band handle anastomat head according to claim 1 is characterized in that, described illuminator (3) is a low angle annular LED light source.

5. the dropped stitch checkout gear of band handle anastomat head according to claim 1, it is characterized in that, described light source shelf (7) is to be made up of five rectangular strips, wherein, article one, be fixed with connecting rod above the rectangular strip, the two limit symmetries a pair of connecting rod of being made up of two rectangular strips that is rotatably connected is formed on four degree of freedom activities all around, outermost two rectangular strips are provided with the M4 screw, are connected by bolt and annular light source lamp.

6. dropped stitch detection method with handle anastomat head is characterized in that it is as follows specifically to detect step:

(1) cmos sensor is gathered original image, and in original image, automatically extract annular related interests zone, the utilization gray scale is cut apart the coarse localization relevant range, utilizes and looks for round operator and radius to carry out match automatically, the annular target area that accurate in locating will detect;

(2) utilize the dual threshold grey level histogram of target area, finish automatic threshold and cut apart, in circle ring area, extract the needle tray region-of-interest, wherein:

Adopt circular expander template in the morphology processing, image can produce a Minkowski increment, effectively removes the noise that the needle tray edge reflections is produced, and its expression formula is as follows:

$A\&CirclePlus;B=\&cup;\{A+b:b\&Element;B\}---\left(1\right)$

Wherein, A is the high-brightness region that extracts, and B is a structural element, select radius be the circle of R as template, translation in the scope of A, the image after obtaining expanding

Image negate after will expanding just can be removed noise when separating high-brightness region, effectively only contained the region-of-interest that detects the needle tray of being concerned about;

(3) image characteristics extraction, utilization be based on the local threshold partitioning algorithm of fuzzy set theory, the region-of-interest image adopted 15 * 15 mean filter smooth template strengthen:

$g(m,n)=\frac{1}{15\&times;15}\mathrm{\&Sigma;\&Sigma;f}(m-k,n-l),(k,l)\&Element;W---\left(2\right)$

F (m, n) and g (m n) is input picture and output image respectively, W be in the position (m, the suitable neighborhood of selecting around pixel n), i.e. pixel in 15 * 15 the spectral window scope, the feature pin that smoothing processing back edge step is bigger, grey scale change is obvious; Comparatively fuzzy base reflection smooth effect relatively a little less than, gray value changes little; Use the soft-threshold of the result of smothing filtering as fuzzy operation again, the degree that belongs to concrete zone according to each pixel is carried out local threshold and is cut apart:

$h(x,y)=\left\{\begin{array}{c}255,f(x,y)\&GreaterEqual;g(x,y)+\mathrm{offset};\\ 0,f(x,y)<g(x,y)+\mathrm{offset};\end{array}\right.---\left(3\right)$

With original image f (x, y) with level and smooth back image g (x, difference y) is greater than the image h (x of offset, y) extract, offset ≠ 0 wherein, choosing of the selection of template and offset has very big motility, can effectively extract the anastomat suture needle of shades of colour model;

(4) adopt area features in the recognition image whether to judge in this needle tray and dropped stitch testing product is divided into i ∈ { I, II| dropped stitch; Pin is arranged } two classes; For the stability of increase differentiating, carry out expansion process earlier, fill the fracture that produces in the cutting procedure, make each feature pin form the connected region of an integral body, the image of each connected region be F (x, y), judge successively according to each connected region area:

$M=\underset{x}{\mathrm{\&Sigma;}}\underset{y}{\mathrm{\&Sigma;}}F(x,y)---\left(4\right)$

M is a cell area, and output suture needle number judges whether product is qualified.

Images