CAJ | 학술논문

为实现含有蔗芽的有效蔗种片段机器智能切断，运用图像处理技术对甘蔗茎节进行识别定位。通过背景转换、灰度级变换、中值滤波和自动阈值获取甘蔗的二值化图像；采用阈值和孔洞填充实现甘蔗区域的分割，通过regionprops 函数测定该区域的质心、倾角和等效长短轴长度；以分割后的甘蔗图像对二值化图像掩膜，得到含有茎节和干扰信息的图像；对该图像进行旋转，计算每列像素值之和，统计分析最大值所在列，并结合质心、等效长短轴得到茎节上下端点坐标；以倾角的度数进行反向旋转，最终得到茎节位置。试验结果表明：甘蔗茎节识别与定位方法处理速度快，茎节识别率高，左右端的定位误差分别小于0．9 mm和2．4 mm。
위실현함유자아적유효자충편단궤기지능절단，운용도상처리기술대감자경절진행식별정위。통과배경전환、회도급변환、중치려파화자동역치획취감자적이치화도상；채용역치화공동전충실현감자구역적분할，통과regionprops 함수측정해구역적질심、경각화등효장단축장도；이분할후적감자도상대이치화도상엄막，득도함유경절화간우신식적도상；대해도상진행선전，계산매렬상소치지화，통계분석최대치소재렬，병결합질심、등효장단축득도경절상하단점좌표；이경각적도수진행반향선전，최종득도경절위치。시험결과표명：감자경절식별여정위방법처리속도쾌，경절식별솔고，좌우단적정위오차분별소우0．9 mm화2．4 mm。
To achieve cut of sugarcane fragments containing internodes by intelligence machine effectively , used image processing techniques to identification and location sugarcane internode .By way of background conversion ,grayscale con-version,median filtering and automated threshold to obtain a binary image of sugarcane ,used thresholds and fill holes to extract the effective area of sugarcane ,measured the centroid ,inclination,equivalent major axis and minor axis length of the region by regionprops function;extracted the interest area of cane by mask the binary image; rotated the image and summation the values for each column of pixels , analysised the position of the maximum values , combined with the cen-troid,equivalent major axis and minor axis to get the 2 coordinates of internode ,reverse rotation the degree of inclination to get internode position finally .Experimental results showed that the method of identification and localization of sugar-cane internode had advantage of processing speedly ,identification rate was high ,the positioning error of left and right ends were within 0.9 mm and 2.4 mm.