US 11,798,185 B2
Image analysis system and image analysis method for obtaining object range
Tung-Yu Wu, Yilan County (TW); Chun-Yen Liao, Taoyuan (TW); Chun-Sheng Wu, Taichung (TW); Kao-Tsair Tsai, Taichung (TW); and Chao-Yi Huang, Taichung (TW)
Assigned to WINBOND ELECTRONICS CORP., Taichung (TW)
Filed by Winbond Electronics Corp., Taichung (TW)
Filed on Mar. 23, 2021, as Appl. No. 17/209,741.
Claims priority of application No. 109110218 (TW), filed on Mar. 26, 2020.
Prior Publication US 2021/0304432 A1, Sep. 30, 2021
Int. Cl. G06T 7/60 (2017.01); G06T 7/11 (2017.01); G06T 5/00 (2006.01); G06T 7/73 (2017.01); G06N 3/04 (2023.01); G06N 3/08 (2023.01)
CPC G06T 7/60 (2013.01) [G06N 3/04 (2013.01); G06N 3/08 (2013.01); G06T 5/002 (2013.01); G06T 7/11 (2017.01); G06T 7/73 (2017.01); G06T 2207/10061 (2013.01); G06T 2207/20021 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30148 (2013.01)] 6 Claims
OG exemplary drawing
 
1. An image analysis system, comprising:
an image capturing device, configured to capture a to-be analyzed image; and
a processor, configured to input the to-be analyzed image into a region-based convolutional neural network (RCNN) model; the region-based convolutional neural network model outputs a masked image; the processor calculates a center of a masked object in the masked image and regards the center as a origin of coordinate, searches for a farthest coordinate point from the origin of coordinate in each of the four quadrants relative to the origin of coordinate, generates an image analysis block for each of the farthest coordinate points, and performs post-processing on the image analysis blocks to obtain an object range;
wherein each of the image analysis blocks comprises a current analysis block, and the processor is further configured to perform a smoothing process on the current analysis block while performing the post-processing;
wherein the processor is further configured to vertically scan each pixel of each of the current analysis blocks while performing the post-processing, when the processor scans a first pixel column of the current analysis block vertically, the processor calculates a first pixel intensity for each pixel in the first pixel column, the first pixel intensities form a first pixel curve, the processor calculates the slope of the first pixel curve to obtain a first slope curve, the processor marks the maximum value of the first slope curve as a marking point the processor is further configured to scan each pixel of the current analysis block vertically to obtain a plurality of marked points, calculate an average intensity value corresponding to the marked points, and multiply the average intensity value by a parameter to obtain a threshold value, after filtering the marked points whose pixel intensity is lower than the threshold value, the remaining marked points are substituted into a linear regression algorithm to obtain a regression line segment.