Curve description and matching method based on gradient order dividing and mapping strategy

• Main Authors: WANG JING, LIU YAN, JIA LI-QIN, LIU HONG-MIN, SI ZENG-YAN, LI GUANG-WU, JIANG GUO-QUAN, WANG ZHI-HENG, HUO ZHAN-QIANG
• Format: Patent
• Language: Chinese; English
• Published: 08.03.2017
• Subjects: CALCULATING; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention relates to a curve description and matching method based on a gradient order dividing and mapping strategy. The method comprises steps: two different images for the same scene are photographed from different angles and are inputted to a computer, a Canny edge detection operator is used for carrying out curve section detection on the two images respectively, a supporting area of the curve is determined, the main direction is calculated, subarea dividing is carried out on the supporting area based on gradient magnitudes, gradient order features of each effective point in the supporting area are calculated, the gradient order features are mapped to a 24-dimensional vector feature, a subarea descriptor and a curve descriptor are calculated, and curve matching is carried out by using the curve descriptor based on a two-way matching rule. The method provided by the invention can overcome the instability problem in the existing gray order dividing-based matching method, and the matching performance is