Research On Image Recognition Technology Of Resolving Ability In Medical Endoscope Quality Detection

The optical performance parameters of medical endoscope include field angle,distortion rate,color reduction,resolving ability and depth-of-field distance.Its quality directly affects the inspection results of the inspected personnel.Aiming at the problem of traditional resolving ability and depth-of-field distance detection methods that traditional methods are affected by human factors,complex to operate and inefficient in detection,this paper introduces image recognition technology to automatically detect the resolving ability and depth-of-field distance of medical hard-tube endoscopes.The main contents of this thesis are mainly as the followings:1.An automatic resolving ability recognition method for medical hard-tube endoscopy is proposed.Firstly,according to the geometric characteristics of the field of view circle,the method locates the center of the air field of view by finding the intersection point of the horizontal and vertical diameters.Then the center of the resolving ability test plate is moved to the center of the field of view.According to the gray scale jump characteristics of the line edge,the coordinates of two pairs of short-line markers in the image of resolving ability test target are obtained,and the rectangular area image surrounded by the line markers is extracted.The extracted image is divided into three sub-images: 1-16?17-24 and 25 level.Three sub-images are segmented into several diamond patterns by using a threshold segmentation algorithm based on prior knowledge for images of medical hard-tube endoscopy resolving ability.Then the improved binarization algorithm based on gray histogram is used to binarize all diamond pattern images.Finally,the line number is calculated by voting and compared with the standard value to determine the resolving ability of endoscope.2.An automatic detection method for the closest field depth of medical hard-tube endoscopy is proposed.Firstly,the center of the line combination unit to be identified is coincided with the center of the field of view.Then four diamond pattern images in the line combination unit are extracted by using the region of interest extraction algorithm based on adaptive threshold.Then the number of lines is calculated by voting and compared with the standard value to determine whether the current line combination unit is successfully identified.If the resolving ability can be recognized correctly,the detection equipment is controlled to move so that the center of the next line combination unit coincides with the center of the field of view to recognize the next level resolving ability.If the recognition fails or is incorrect,the unit number of the last line combination unit is output.So then,it can complete the automatic detection of the depth of field of the medical hard-tube endoscope.3.Experimental verification.Three different specifications of medical hard-tube endoscopes were used to test the above methods to verify the effectiveness of this method.The resolving ability recognition method of medical hard-tube endoscopy presented in this paper can be successfully identified in 20 experiments with 4 mm diameter,30 apparent angle and 10 mm diameter,0 apparent angle.Only one of the 20 experiments is different from the results of manual identification and deviation is only one order.In 10 mm diameter and 30 apparent angle were performed with medical hard-tube endoscopy,only two of the 10 experiments results about manual recognition are different and only differ by one level.In 20 experiments of medical hard-tube endoscopy with 10 mm diameter,30 apparent angle and 10 mm diameter and 0 apparent angle,the field depth distance detection method of medical hard endoscopy presented in this paper has higher than 80% of the central resolving ability at the working distance,and only one time is lower than the result of manual identification.In the 10 experiments of medical hard-tube endoscopy with 4 mm diameter and 30 apparent angle,all of them can be successfully identified,and only two of which are different from the results of manual recognition and deviation is only one order.Two conclusions can be drawn from the experimental results.The first is that the resolving ability recognition method proposed in this paper can realize the automatic recognition of the resolving ability of medical hard endoscopes.The accuracy rate is 80%.The second is that the depth-of-field distance detection method of medical hard-tube endoscopes proposed in this paper can realize the automatic detection of the depth-of-field distance of medical hard endoscopes.The results of detection are more than 80% of the resolving ability of the central angle.The method of medical endoscopy's resolving ability recognition and the method of medical endoscopy's depth-of-field closest distance detection proposed in this paper meet the requirements of detection accuracy.