create_uncalib_descriptor_modelT_create_uncalib_descriptor_modelCreateUncalibDescriptorModelCreateUncalibDescriptorModelcreate_uncalib_descriptor_model创建未标定描述符模型(算子)
名称
create_uncalib_descriptor_modelT_create_uncalib_descriptor_modelCreateUncalibDescriptorModelCreateUncalibDescriptorModelcreate_uncalib_descriptor_model — 为兴趣点匹配准备描述符模型。
签名
void CreateUncalibDescriptorModel(const HObject& Template, const HTuple& DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, const HTuple& Seed, HTuple* ModelID)
void HDescriptorModel::HDescriptorModel(const HImage& Template, const HString& DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed)
void HDescriptorModel::HDescriptorModel(const HImage& Template, const char* DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed)
void HDescriptorModel::HDescriptorModel(const HImage& Template, const wchar_t* DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed)
(
Windows only)
void HDescriptorModel::CreateUncalibDescriptorModel(const HImage& Template, const HString& DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed)
void HDescriptorModel::CreateUncalibDescriptorModel(const HImage& Template, const char* DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed)
void HDescriptorModel::CreateUncalibDescriptorModel(const HImage& Template, const wchar_t* DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed)
(
Windows only)
HDescriptorModel HImage::CreateUncalibDescriptorModel(const HString& DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed) const
HDescriptorModel HImage::CreateUncalibDescriptorModel(const char* DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed) const
HDescriptorModel HImage::CreateUncalibDescriptorModel(const wchar_t* DetectorType, const HTuple& DetectorParamName, const HTuple& DetectorParamValue, const HTuple& DescriptorParamName, const HTuple& DescriptorParamValue, Hlong Seed) const
(
Windows only)
static void HOperatorSet.CreateUncalibDescriptorModel(HObject template, HTuple detectorType, HTuple detectorParamName, HTuple detectorParamValue, HTuple descriptorParamName, HTuple descriptorParamValue, HTuple seed, out HTuple modelID)
public HDescriptorModel(HImage template, string detectorType, HTuple detectorParamName, HTuple detectorParamValue, HTuple descriptorParamName, HTuple descriptorParamValue, int seed)
void HDescriptorModel.CreateUncalibDescriptorModel(HImage template, string detectorType, HTuple detectorParamName, HTuple detectorParamValue, HTuple descriptorParamName, HTuple descriptorParamValue, int seed)
HDescriptorModel HImage.CreateUncalibDescriptorModel(string detectorType, HTuple detectorParamName, HTuple detectorParamValue, HTuple descriptorParamName, HTuple descriptorParamValue, int seed)
def create_uncalib_descriptor_model(template: HObject, detector_type: str, detector_param_name: Sequence[str], detector_param_value: Sequence[Union[int, float, str]], descriptor_param_name: Sequence[str], descriptor_param_value: Sequence[Union[int, float, str]], seed: int) -> HHandle
描述
算子 create_uncalib_descriptor_modelcreate_uncalib_descriptor_modelCreateUncalibDescriptorModelCreateUncalibDescriptorModelCreateUncalibDescriptorModelcreate_uncalib_descriptor_model 用于构建图像
TemplateTemplateTemplateTemplatetemplatetemplate 中指定区域的描述符模型,该模型可用于基于描述符的匹配。通过随后调用 find_uncalib_descriptor_modelfind_uncalib_descriptor_modelFindUncalibDescriptorModelFindUncalibDescriptorModelFindUncalibDescriptorModelfind_uncalib_descriptor_model,可以获得从 TemplateTemplateTemplateTemplatetemplatetemplate 到搜索图像的投影二维变换(单应性变换)。TemplateTemplateTemplateTemplatetemplatetemplate 中域区域的重心被用作模型的原点。与 create_calib_descriptor_modelcreate_calib_descriptor_modelCreateCalibDescriptorModelCreateCalibDescriptorModelCreateCalibDescriptorModelcreate_calib_descriptor_model 不同,此处无需校准相机,因此后续配准的结果为二维投影。请注意,TemplateTemplateTemplateTemplatetemplatetemplate 图像中可见的对象部分必须是平面的。
描述符模型描述了一组兴趣点。它存储了这些兴趣点的位置及其局部灰度邻域的判别性描述。兴趣点的提取由 DetectorTypeDetectorTypeDetectorTypeDetectorTypedetectorTypedetector_type、DetectorParamNameDetectorParamNameDetectorParamNameDetectorParamNamedetectorParamNamedetector_param_name 和 DetectorParamValueDetectorParamValueDetectorParamValueDetectorParamValuedetectorParamValuedetector_param_value 参数化。兴趣点周围的相应描述符由 DescriptorParamNameDescriptorParamNameDescriptorParamNameDescriptorParamNamedescriptorParamNamedescriptor_param_name 和 DescriptorParamValueDescriptorParamValueDescriptorParamValueDescriptorParamValuedescriptorParamValuedescriptor_param_value 参数化。参数 SeedSeedSeedSeedseedseed 用于初始化随返回的
ModelIDModelIDModelIDModelIDmodelIDmodel_id 是对生成的描述符模型的引用。机数生成器,该生成器在基于 随机蕨randomized ferns 实现的描述符构建过程中会被用到。该模型可用于通过
find_uncalib_descriptor_modelfind_uncalib_descriptor_modelFindUncalibDescriptorModelFindUncalibDescriptorModelFindUncalibDescriptorModelfind_uncalib_descriptor_model 高效检测已学习模板的实例,从而实现模型与搜索图像之间的透视变换。由于基于描述符的匹配依赖于稳定且具有区分度的兴趣点,因此待检测的对象必须具有纹理,但纹理不应呈重复模式。
检测器参数
如前所述,检测器用于从图像中提取稳定的兴趣点。通过参数
DetectorTypeDetectorTypeDetectorTypeDetectorTypedetectorTypedetector_type,可以选择要使用的兴趣点算子。目前支持 points_lepetitpoints_lepetitPointsLepetitPointsLepetitPointsLepetitpoints_lepetit、points_harrispoints_harrisPointsHarrisPointsHarrisPointsHarrispoints_harris 及其二项式近似
points_harris_binomialpoints_harris_binomialPointsHarrisBinomialPointsHarrisBinomialPointsHarrisBinomialpoints_harris_binomial('lepetit'、'harris'、'harris_binomial')。若模板或搜索图像对比度较低,应选用 Harris 点算子之一。根据所选的 DetectorTypeDetectorTypeDetectorTypeDetectorTypedetectorTypedetector_type,可在 DetectorParamNameDetectorParamNameDetectorParamNameDetectorParamNamedetectorParamNamedetector_param_name 和 DetectorParamValueDetectorParamValueDetectorParamValueDetectorParamValuedetectorParamValuedetector_param_value 中设置相应的参数名称和数值。
DetectorParamNameDetectorParamNameDetectorParamNameDetectorParamNamedetectorParamNamedetector_param_name 的可用参数名称及其默认值如下:
- 'lepetit'"lepetit""lepetit""lepetit""lepetit""lepetit":
-
['radius'"radius""radius""radius""radius""radius", 'check_neighbor'"check_neighbor""check_neighbor""check_neighbor""check_neighbor""check_neighbor",
'min_check_neighbor_diff'"min_check_neighbor_diff""min_check_neighbor_diff""min_check_neighbor_diff""min_check_neighbor_diff""min_check_neighbor_diff", 'min_score'"min_score""min_score""min_score""min_score""min_score",
'subpix'"subpix""subpix""subpix""subpix""subpix"]
[ 3, 1, 15, 30, 'interpolation'][ 3, 1, 15, 30, "interpolation"][ 3, 1, 15, 30, "interpolation"][ 3, 1, 15, 30, "interpolation"][ 3, 1, 15, 30, "interpolation"][ 3, 1, 15, 30, "interpolation"]
- 'harris'"harris""harris""harris""harris""harris":
-
['sigma_grad'"sigma_grad""sigma_grad""sigma_grad""sigma_grad""sigma_grad", 'sigma_smooth'"sigma_smooth""sigma_smooth""sigma_smooth""sigma_smooth""sigma_smooth", 'alpha'"alpha""alpha""alpha""alpha""alpha",
'threshold'"threshold""threshold""threshold""threshold""threshold"]
[0.7, 2.0, 0.08, 1000]
- 'harris_binomial'"harris_binomial""harris_binomial""harris_binomial""harris_binomial""harris_binomial":
-
['mask_size_grd'"mask_size_grd""mask_size_grd""mask_size_grd""mask_size_grd""mask_size_grd", 'mask_size_smooth'"mask_size_smooth""mask_size_smooth""mask_size_smooth""mask_size_smooth""mask_size_smooth",
'alpha'"alpha""alpha""alpha""alpha""alpha", 'threshold'"threshold""threshold""threshold""threshold""threshold", 'subpix'"subpix""subpix""subpix""subpix""subpix"]
[5, 15, 0.08, 1000, 'on'][5, 15, 0.08, 1000, "on"][5, 15, 0.08, 1000, "on"][5, 15, 0.08, 1000, "on"][5, 15, 0.08, 1000, "on"][5, 15, 0.08, 1000, "on"]
有关这些参数含义的更多详细信息,请分别参阅 points_harrispoints_harrisPointsHarrisPointsHarrisPointsHarrispoints_harris、points_harris_binomialpoints_harris_binomialPointsHarrisBinomialPointsHarrisBinomialPointsHarrisBinomialpoints_harris_binomial 和 points_lepetitpoints_lepetitPointsLepetitPointsLepetitPointsLepetitpoints_lepetit 的说明。如果传递的是空元组,或者在 DetectorParamNameDetectorParamNameDetectorParamNameDetectorParamNamedetectorParamNamedetector_param_name 中未提供参数,则采用上述默认值。
调整算子参数时,应确保提取出 50 至 450 个特征点(具体数量取决于 TemplateTemplateTemplateTemplatetemplatetemplate 的纹理和尺寸),且这些特征点在模板的 ROI 上均匀分布。因此,建议预先运行所选的点算子,并通过
gen_cross_contour_xldgen_cross_contour_xldGenCrossContourXldGenCrossContourXldGenCrossContourXldgen_cross_contour_xld 可视化结果。在大多数情况下,使用默认设置即可满足需求。
描述符参数
点描述符是一种分类器,它为兴趣点构建其灰度邻域的特征描述。目前,该描述符采用所谓的 随机蕨randomized ferns 算法实现,该算法通过学习兴趣点周围区域中 随机 选定位置的像素对之间灰度差的极性来工作。该描述符随后将在 find_uncalib_descriptor_modelfind_uncalib_descriptor_modelFindUncalibDescriptorModelFindUncalibDescriptorModelFindUncalibDescriptorModelfind_uncalib_descriptor_model 中用于对搜索图像中的兴趣点进行分类,换言之:用于识别(匹配)搜索图像中的潜在模型点。
描述子只需存储投影 稳定 的兴趣点(这些点将在模板的多个投影视图中出现)。为了评估兴趣点的 稳定性,会进行一次模拟:对 TemplateTemplateTemplateTemplatetemplatetemplate 进行多次仿射变换,并在大多数视图中都能提取出的点被视为 稳定点。这些仿射变换很好地近似了兴趣点局部邻域内的投影变换。
可以通过
DescriptorParamNameDescriptorParamNameDescriptorParamNameDescriptorParamNamedescriptorParamNamedescriptor_param_name 和 DescriptorParamValueDescriptorParamValueDescriptorParamValueDescriptorParamValuedescriptorParamValuedescriptor_param_value 设置以下描述符参数:
描述符大小参数:
- 'depth'"depth""depth""depth""depth""depth":
-
分类蕨的深度。更深的随机蕨能更好地区分兴趣点。然而,蕨的内存需求会按 'depth'"depth""depth""depth""depth""depth" 的 2 的幂次方增长。典型取值范围为 [5 .. 11],默认值为 11。
- 'number_ferns'"number_ferns""number_ferns""number_ferns""number_ferns""number_ferns":
-
使用的蕨结构数量。使用更多的蕨会提高识别稳健性,但也会增加匹配的运行时间。如果需要使描述符的内存占用较小,应使用大量深度较小的蕨(例如,'number_ferns'"number_ferns""number_ferns""number_ferns""number_ferns""number_ferns"=150,'depth'"depth""depth""depth""depth""depth"=5)。若更重视检测速度,则应使用较少数量但深度较大的蕨(例如,'number_ferns'"number_ferns""number_ferns""number_ferns""number_ferns""number_ferns"=10,'depth'"depth""depth""depth""depth""depth"=11)。典型取值范围为 [1 .. 150],默认值为 30。
- 'patch_size'"patch_size""patch_size""patch_size""patch_size""patch_size":
描述兴趣点的二次邻域边长。该参数值过大可能会影响运行时间。典型取值范围为 [15 .. 33],默认值为 17。
综上所述,参数 'depth'"depth""depth""depth""depth""depth"、'number_ferns'"number_ferns""number_ferns""number_ferns""number_ferns""number_ferns" 和
'patch_size'"patch_size""patch_size""patch_size""patch_size""patch_size" 允许对检测的稳健性、速度和内存消耗进行直观控制。
模拟参数:
- 'tilt'"tilt""tilt""tilt""tilt""tilt":
-
控制在模拟阶段是否启用投影变换。启用后,模型的稳健性将得到提升,能够检测到倾斜角度更大的对象;关闭后,训练时间可显著缩短,且模型仍能识别具有投影不变性的对象。取值范围为 ['on'"on""on""on""on""on", 'off'"off""off""off""off""off"],默认值为 'on'"on""on""on""on""on"。
- 'min_rot'"min_rot""min_rot""min_rot""min_rot""min_rot":
-
TemplateTemplateTemplateTemplatetemplatetemplate 法线向量周围的最小旋转角度。典型取值范围为 [-180 .. 0],默认值为 -180。
- 'max_rot'"max_rot""max_rot""max_rot""max_rot""max_rot":
-
TemplateTemplateTemplateTemplatetemplatetemplate 法线向量周围的最大旋转角度。典型取值范围为 [0 .. 180],默认值为 180。
- 'min_scale'"min_scale""min_scale""min_scale""min_scale""min_scale":
-
TemplateTemplateTemplateTemplatetemplatetemplate 的最小缩放比例。典型取值范围为 [0.1 .. 1.0],默认值为 0.5。
- 'max_scale'"max_scale""max_scale""max_scale""max_scale""max_scale":
TemplateTemplateTemplateTemplatetemplatetemplate 的最大缩放比例。典型取值范围为 [1.0 .. 3.5],默认值为 1.4。
参数 'min_rot'"min_rot""min_rot""min_rot""min_rot""min_rot"、'max_rot'"max_rot""max_rot""max_rot""max_rot""max_rot"、'min_scale'"min_scale""min_scale""min_scale""min_scale""min_scale" 和 'max_scale'"max_scale""max_scale""max_scale""max_scale""max_scale" 允许手动设置用于训练描述符的模板的仿射变换视图。设置这些参数有助于缩短训练时间,尤其是与 'tilt'"tilt""tilt""tilt""tilt""tilt" 参数结合使用时。请注意,这些参数会直接影响
find_uncalib_descriptor_modelfind_uncalib_descriptor_modelFindUncalibDescriptorModelFindUncalibDescriptorModelFindUncalibDescriptorModelfind_uncalib_descriptor_model 的结果,因此必须谨慎设置。例如,如果旋转范围被限制在
'min_rot'"min_rot""min_rot""min_rot""min_rot""min_rot" = -10 至 'max_rot'"max_rot""max_rot""max_rot""max_rot""max_rot" = 10 之间,则无法找到旋转角度超出该范围的 TemplateTemplateTemplateTemplatetemplatetemplate 视图。受限的训练范围意味着找到 TemplateTemplateTemplateTemplatetemplatetemplate 所需的蕨数量和深度较少。在这种情况下,可以进一步减少蕨的数量和深度,从而优化模型。
备注
请注意,根据您的硬件配置不同,训练随机蕨模型的处理时间可能在几秒到几分钟之间。因此,训练好的模型可以使用
write_descriptor_modelwrite_descriptor_modelWriteDescriptorModelWriteDescriptorModelWriteDescriptorModelwrite_descriptor_model 和 read_descriptor_modelread_descriptor_modelReadDescriptorModelReadDescriptorModelReadDescriptorModelread_descriptor_model 进行保存和加载。
最终描述符点的参数和位置可以通过 get_descriptor_model_paramsget_descriptor_model_paramsGetDescriptorModelParamsGetDescriptorModelParamsGetDescriptorModelParamsget_descriptor_model_params 和 get_descriptor_model_pointsget_descriptor_model_pointsGetDescriptorModelPointsGetDescriptorModelPointsGetDescriptorModelPointsget_descriptor_model_points 获取。
create_uncalib_descriptor_modelcreate_uncalib_descriptor_modelCreateUncalibDescriptorModelCreateUncalibDescriptorModelCreateUncalibDescriptorModelcreate_uncalib_descriptor_model 会存储检测器类型、检测器参数和描述符参数,这些参数将在后续每次调用 find_uncalib_descriptor_modelfind_uncalib_descriptor_modelFindUncalibDescriptorModelFindUncalibDescriptorModelFindUncalibDescriptorModelfind_uncalib_descriptor_model 时使用。模型的参考点(原点)是模板 ROI 的质心。其坐标可通过
set_descriptor_model_originset_descriptor_model_originSetDescriptorModelOriginSetDescriptorModelOriginSetDescriptorModelOriginset_descriptor_model_origin 进行修改。
执行信息
- 多线程类型:可重入(与非独占算子并行运行)。
- 多线程作用域:全局(可从任何线程调用)。
- 未采用并行化处理。
此算子返回一个句柄。请注意,即使该句柄被用作特定算子的输入参数,这些算子仍可能改变此句柄类型的实例状态。
参数
TemplateTemplateTemplateTemplatetemplatetemplate (输入对象) singlechannelimage → objectHImageHObjectHImageHobject (byte / uint2)
输入图像,其域将用于创建模型。
DetectorTypeDetectorTypeDetectorTypeDetectorTypedetectorTypedetector_type (输入控制) string → HTuplestrHTupleHtuple (string) (string) (HString) (char*)
检测器的类型。
默认值:
'lepetit'
"lepetit"
"lepetit"
"lepetit"
"lepetit"
"lepetit"
值列表:
'harris'"harris""harris""harris""harris""harris", 'harris_binomial'"harris_binomial""harris_binomial""harris_binomial""harris_binomial""harris_binomial", 'lepetit'"lepetit""lepetit""lepetit""lepetit""lepetit"
DetectorParamNameDetectorParamNameDetectorParamNameDetectorParamNamedetectorParamNamedetector_param_name (输入控制) attribute.name-array → HTupleSequence[str]HTupleHtuple (string) (string) (HString) (char*)
检测器的参数名称。
默认值:
[]
值列表:
'alpha'"alpha""alpha""alpha""alpha""alpha", 'check_neighbor'"check_neighbor""check_neighbor""check_neighbor""check_neighbor""check_neighbor", 'mask_size_grd'"mask_size_grd""mask_size_grd""mask_size_grd""mask_size_grd""mask_size_grd", 'mask_size_smooth'"mask_size_smooth""mask_size_smooth""mask_size_smooth""mask_size_smooth""mask_size_smooth", 'min_check_neighbor_diff'"min_check_neighbor_diff""min_check_neighbor_diff""min_check_neighbor_diff""min_check_neighbor_diff""min_check_neighbor_diff", 'min_score'"min_score""min_score""min_score""min_score""min_score", 'radius'"radius""radius""radius""radius""radius", 'sigma_grad'"sigma_grad""sigma_grad""sigma_grad""sigma_grad""sigma_grad", 'sigma_smooth'"sigma_smooth""sigma_smooth""sigma_smooth""sigma_smooth""sigma_smooth", 'subpix'"subpix""subpix""subpix""subpix""subpix", 'threshold'"threshold""threshold""threshold""threshold""threshold"
DetectorParamValueDetectorParamValueDetectorParamValueDetectorParamValuedetectorParamValuedetector_param_value (输入控制) attribute.value-array → HTupleSequence[Union[int, float, str]]HTupleHtuple (integer / real / string) (int / long / double / string) (Hlong / double / HString) (Hlong / double / char*)
检测器的参数值。
默认值:
[]
建议值:
0.08, 1, 1.2, 3, 15, 30, 1000, 'on'"on""on""on""on""on", 'off'"off""off""off""off""off", 'none'"none""none""none""none""none", 'interpolation'"interpolation""interpolation""interpolation""interpolation""interpolation"
DescriptorParamNameDescriptorParamNameDescriptorParamNameDescriptorParamNamedescriptorParamNamedescriptor_param_name (输入控制) attribute.name-array → HTupleSequence[str]HTupleHtuple (string) (string) (HString) (char*)
描述符的参数名称。
默认值:
[]
值列表:
'depth'"depth""depth""depth""depth""depth", 'max_rot'"max_rot""max_rot""max_rot""max_rot""max_rot", 'max_scale'"max_scale""max_scale""max_scale""max_scale""max_scale", 'min_rot'"min_rot""min_rot""min_rot""min_rot""min_rot", 'min_scale'"min_scale""min_scale""min_scale""min_scale""min_scale", 'number_ferns'"number_ferns""number_ferns""number_ferns""number_ferns""number_ferns", 'patch_size'"patch_size""patch_size""patch_size""patch_size""patch_size", 'tilt'"tilt""tilt""tilt""tilt""tilt"
DescriptorParamValueDescriptorParamValueDescriptorParamValueDescriptorParamValuedescriptorParamValuedescriptor_param_value (输入控制) attribute.value-array → HTupleSequence[Union[int, float, str]]HTupleHtuple (integer / real / string) (int / long / double / string) (Hlong / double / HString) (Hlong / double / char*)
描述符的参数值。
默认值:
[]
建议值:
0.5, 1.4, 11, 21, 30, -180, 180, 'on'"on""on""on""on""on", 'off'"off""off""off""off""off"
SeedSeedSeedSeedseedseed (输入控制) integer → HTupleintHTupleHtuple (integer) (int / long) (Hlong) (Hlong)
随机数生成器的种子。
默认值:
42
ModelIDModelIDModelIDModelIDmodelIDmodel_id (输出控制) descriptor_model → HDescriptorModel, HTupleHHandleHTupleHtuple (handle) (IntPtr) (HHandle) (handle)
描述符模型的句柄。
可能的前趋
points_lepetitpoints_lepetitPointsLepetitPointsLepetitPointsLepetitpoints_lepetit,
points_harrispoints_harrisPointsHarrisPointsHarrisPointsHarrispoints_harris,
reduce_domainreduce_domainReduceDomainReduceDomainReduceDomainreduce_domain
可能的后继
get_descriptor_model_paramsget_descriptor_model_paramsGetDescriptorModelParamsGetDescriptorModelParamsGetDescriptorModelParamsget_descriptor_model_params,
find_uncalib_descriptor_modelfind_uncalib_descriptor_modelFindUncalibDescriptorModelFindUncalibDescriptorModelFindUncalibDescriptorModelfind_uncalib_descriptor_model
另见
get_descriptor_model_paramsget_descriptor_model_paramsGetDescriptorModelParamsGetDescriptorModelParamsGetDescriptorModelParamsget_descriptor_model_params,
find_uncalib_descriptor_modelfind_uncalib_descriptor_modelFindUncalibDescriptorModelFindUncalibDescriptorModelFindUncalibDescriptorModelfind_uncalib_descriptor_model
参考文献
V. Lepetit and P. Fua: “Keypoint Recognition using Randomized Trees.“
IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 28,
Nr. 9, pp. 1465-1479, 2006.
M. Ozuysal, P. Fua, and V. Lepetit: “Fast Keypoint Recognition in Ten Lines
of Code.“
In Proceedings of Conference on Computer Vision and Pattern Recognition,
2007.
模块
匹配