农机化研究
農機化研究
농궤화연구
JOURNAL OF AGRICULTURAL MECHANIZATION RESEARCH
2014年
3期
28-31
,共4页
分级方法%红枣%图像处理%模型
分級方法%紅棘%圖像處理%模型
분급방법%홍조%도상처리%모형
image processing%red jujubes%machine vision%model
为了提高红枣商品品质,提出了红枣静态图像采集分级方法。通过比较不同光照条件、颜色背景下取得的红枣图片,研究基于机器视觉红枣大小的分级检测方法。采集红枣图像和图像预处理,选取最佳图像阈值,提取特征尺寸,建立特征尺寸与质量之间的对应关系;建立红枣的机测尺寸与实际尺寸、机测尺寸与红枣质量、实际尺寸与质量之间的不同关系模型。选择最贴近的二维分级模型:红枣机测实面积与质量的一次线性回归模型,通过对机测面积线性方程回归分析,线性模型的确定系数R2=0.8418,并对研究结果进行验证,为进一步开展在线机器视觉红枣品质分级奠定了基础。
為瞭提高紅棘商品品質,提齣瞭紅棘靜態圖像採集分級方法。通過比較不同光照條件、顏色揹景下取得的紅棘圖片,研究基于機器視覺紅棘大小的分級檢測方法。採集紅棘圖像和圖像預處理,選取最佳圖像閾值,提取特徵呎吋,建立特徵呎吋與質量之間的對應關繫;建立紅棘的機測呎吋與實際呎吋、機測呎吋與紅棘質量、實際呎吋與質量之間的不同關繫模型。選擇最貼近的二維分級模型:紅棘機測實麵積與質量的一次線性迴歸模型,通過對機測麵積線性方程迴歸分析,線性模型的確定繫數R2=0.8418,併對研究結果進行驗證,為進一步開展在線機器視覺紅棘品質分級奠定瞭基礎。
위료제고홍조상품품질,제출료홍조정태도상채집분급방법。통과비교불동광조조건、안색배경하취득적홍조도편,연구기우궤기시각홍조대소적분급검측방법。채집홍조도상화도상예처리,선취최가도상역치,제취특정척촌,건립특정척촌여질량지간적대응관계;건립홍조적궤측척촌여실제척촌、궤측척촌여홍조질량、실제척촌여질량지간적불동관계모형。선택최첩근적이유분급모형:홍조궤측실면적여질량적일차선성회귀모형,통과대궤측면적선성방정회귀분석,선성모형적학정계수R2=0.8418,병대연구결과진행험증,위진일보개전재선궤기시각홍조품질분급전정료기출。
In order to improve the quality of jujube commodity , jujube static image acquisition grading method .Jujube pictures obtained by comparing different light shining under the conditions of the study a hierarchical detection method based on machine vision jujube size .Acquisition jujube image and image preprocessing threshold select the best image , extracting the feature size and establish the correspondence between the characteristic dimensions and weight .Establish a different relationship model between the dates in the size of the machine measured actual size , the machine measured di-mensions and jujube weight , the actual size and weight , and to select the most close to the two-dimensional hierarchical model:Jujube machine measuring real size and weight of a quadratic curve regression model ,on the machine side of the area of linear equations regression analysis , curve model coefficient of determination R2=0 .847 9 , and validate find-ings .Determined machine planimetry jujube size fractionation method of .Laid the foundation for further development of online machine vision jujube quality grading .