Research On Ice Spoon Splitting Detection Method

Ice spoon is a tool for eating ice cream.It will be cracked when cut by the machine during production.If the ice spoon with a crack enters the market,it will lead to scratches on the user’s tongue,thus causing certain potential safety hazards.Therefore,strict quality inspection is required before the ice scoop is put into use.According to the literature search,the current split detection of ice scoop can not meet the requirements of enterprises for the quality of ice scoop.According to the characteristics of ice scoop splitting on the gray distribution curve,this thesis realizes the detection of ice scoop splitting.The main research contents of this thesis are as follows:For the positioning of the front and side of the ice scoop,the positioning is mainly carried out by extracting the front waist boundary and the side head boundary of the ice scoop.For detection of frontal waist boundary of ice scoop,this thesis extracts the boundary based on the amplitude characteristics of the transition segment of frontal boundary of ice scoop.For the detection of the side head boundary of the ice scoop,this thesis uses image segmentation method based on gray level threshold to extract the side head boundary,and uses affine transformation to correct the direction of the side head of the ice scoop.The front of the scoop splits into waist splits,wide splits and narrow splits.Lumbar splitting is a splitting located at the boundary of an ice scoop.In this thesis,the detection of lumbar splitting is achieved by extracting the minimum gray scale points or gradient changes within the transition line segment.For wide splitting,this thesis achieves wide splitting detection by extracting the amplitude characteristics of the combined concave segments on the gray distribution curve.For dark mineral lines on the front of the ice scoop that resemble the widesplitting characteristics,the distinction is made by statistic of the polarity of the amplitudes shown in the concave segments at the wide-splitting and dark-splitting mineral lines.For narrow splitting,this thesis uses the least squares method to iteratively fit the gray distribution curve on the front of the ice scoop,and compares it with the original gray distribution curve.The detection of narrow splitting is achieved by extracting the difference feature of the two curves.For light-colored mineral lines with similar narrow splitting characteristics on the front of the ice scoop,the difference is made by counting the gray-scale mutations that the narrow splitting and lightcolored mineral lines show on the concave segments.Finally,by testing the pictures in the gallery,the accuracy of front splitting detection in this thesis is higher.For the side head splitting of ice scoop,this thesis eliminates the influence of uneven illumination on the side head by gray correction,and detects the splitting by extracting the amplitude characteristics of concave segments of the side head gray distribution curve after gray correction.For splitting with a higher vertical degree of the head on the side of the ice scoop,a method of calculating the gray mean of the vertical direction of the head area on the side of the ice scoop before gray scale correction is presented,which enhances the amplitude characteristics of the splitting position.This method can effectively avoid the low contrast caused by illumination and the unclear splitting characteristics caused by complex background in the side image of ice scoop,and reduce the rate of miss and mistake detection.