In this paper, we propose a simple and low complexity pulse peak detection algorithm using cascaded recursive digital filters and a slope sum function (SSF) with an adaptive thresholding scheme. The algorithm first eliminates noises in the photoplethysmogram (PPG) using the cascaded lowpass and highpass digital filters. The filters have been designed with 3-dB cutoff frequencies of 11 Hz and 0.5 Hz, respectively. The filtered PPG signal is then transformed by the SSF. The SSF simplifies detecting the pulse peaks by enhancing the upslope of the PPG signal and suppressing the remainder. A threshold for identifying SSF peaks is updated using the median filter with an order of 5. This update makes the threshold adaptive to variations of SSF heights. The detected SSF peaks localize ranges for pulse peak detection. Finally, the pulse peak is identified by picking the local maxima within the range from an onset index of the SSF signal to the following zero index. In order to cope with over-detected and missed information, the proposed algorithm employs knowledge-based rules as post-processing. The algorithm is tested on a database where PPG waveforms are collected from 127 subjects. The results are promising, suggesting that the method provides simpler but accurate pulse peak detection in real applications.