Masato Kazui

Incoherent motion detection using a time-series Gram matrix feature

This paper proposes a new method for incoherent motion recognition from video sequences. We use time-series spatio-temporal intensity gradients within a space-time patch. Using a global space-time patch, we found that the gradient feature allows us to distinguish an incoherent motion from a coherent motion without segmentation. Furthermore the algorithm can run in real time even on an embedded device. In this paper, we verify motion recognition performance for actions which we consider coherent (walk/run) and incoherent (turn/squat/inverse walk). To identify the multiple motion classes, we use linear discriminant analysis and the KNN method. As a result, Our method can distinguish multiple-class motion patterns with a detection rate of about 80%. Also the detection rule of incoherent motions is 100% with a false positive rate of less than 10%.

Robust object detection based on radial reach correlation and adaptive background estimation for real-time video surveillance systems

A method of real-time object detection for video surveillance systems has been developed. The method aims to realize robust object detection by using Radial Reach Correlation (RRC). We also apply a statistical background estimation to cope with dynamic and complex environments. The computational cost of RRC is higher than the simple subtraction method and the background estimation method based on statistical approach needs large memory. It is necessary to reduce the calculation cost in order to apply to an embedded image processing device. Our method is composed of two techniques: fast RRC algorithm and background estimation based on statistical approach with cumulative averaging process. As a result, without deterioration in detection accuracy, the processing time of object detection can be decreased to about 1/4 in comparison with normal RRC.

3D Camera Calibration Using Gray Code Patterns

This paper presents a new calibration method that uses gray code patterns as a planar chart displayed on an LCD and is based on a parametrical algorithm for stereo cameras. It enables to get high accuracy with easy operation and is well suited for stereo cameras with zoom lenses.

Motion detection with camera shake

A method for detecting an objects motion in images that suffer from camera shake or images with camera egomotion is proposed. This approach is based on edge orientation codes and on the entropy calculated from a histogram of the edge orientation codes. Here, entropy is extended to spatio-temporal entropy. We consider that the spatio-temporal entropy calculated from time-series orientation codes can represent motion complexity, e.g., the motion of a pedestrian. Our method can reject false positives caused by camera shake or background motion. Before the motion filtering, object candidates are detected by a frame-subtraction-based method. After the filtering, over-detected candidates are evaluated using the spatio-temporal entropy, and false positives are then rejected by a threshold. This method could reject 79 to 96 [%] of all false positives in road roller and escalator scenes. The motion filtering decreased the detection rate somewhat because of motion coherency or small apparent motion of a target. In such cases, we need to introduce a tracking method such as Particle Filter or Mean Shift Tracker. The running speed of our method is 32 to 46 ms per frame with a 160×120 pixel image on an Intel Pentium 4 CPU at 2.8 GHz. We think that this is fast enough for real-time detection. In addition, our method can be used as pre-processing for classifiers based on support vector machines or Boosting.

Rapid object candidate detection using increment sign correlation

We develop a rapid object-candidates detector using Increment Sign Correlation (ISC). Our method aims to detect candidates of objects such as people or vehicles in real time using ISC and a simple shape model. Our method is similar to Generalized Hough Transform (GHT). However we modify its voting process. We use ISC for detecting object candidates instead of the shape voting done by GHT. ISC is robust against shading and low image contrast due to lighting changes because Increment Sign (IS) is insensitive to a perturbation of direction of intensity gradient. The computational cost of IS is lower than that of the gradient also. From the results of our experiment, our detector can run with a 320×240 pixel image within 32 milliseconds on a Pentium 4 processor at 2.8 GHz. Given the initial template size of 10×20 pixels, the number of candidates decreases from 170,196 sub-windows in a 320×240 pixel image to 400 at most with the miss rate of 0.2 %. The detection rate is enough for more precise detectors which need to use richer image features. The experimental results using real image sequences are reported.